fix compilation warning

bump version to v2.1.0
Merge pull request 'Merge changes from v1.x into v2.x' (#1161 ) from sync-v2-to-v1 into main-v2
2026-08-13 23:56:53 +00:00 · 2025-09-15 15:52:53 +02:00 · 2025-09-15 15:50:02 +02:00 · 2025-09-15 13:24:20 +00:00 · 2025-09-14 21:04:04 +02:00 · 2025-09-14 18:28:01 +00:00
142 changed files with 3772 additions and 1249 deletions
@@ -1,3 +1,6 @@
+labels:
+  nix: "enabled"
+
 when:
  event:
  - push
@@ -28,6 +31,11 @@ steps:
    commands:
      - nix-build -j4 --attr flakePackages.tests-sqlite

+  - name: unit + func tests (fjall)
+    image: nixpkgs/nix:nixos-22.05
+    commands:
+      - nix-build -j4 --attr flakePackages.tests-fjall
+
  - name: integration tests
    image: nixpkgs/nix:nixos-22.05
    commands:
@@ -1,3 +1,6 @@
+labels:
+  nix: "enabled"
+
 when:
  event:
  - deployment
@@ -1,3 +1,6 @@
+labels:
+  nix: "enabled"
+
 when:
  event:
  - deployment
@@ -556,7 +556,7 @@ checksum = "b62ddb9cb1ec0a098ad4bbf9344d0713fa193ae1a80af55febcff2627b6a00c1"
 dependencies = [
 "getrandom 0.2.15",
 "instant",
- "rand",
+ "rand 0.8.5",
 ]

 [[package]]
@@ -687,6 +687,12 @@ version = "1.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2d2c12f985c78475a6b8d629afd0c360260ef34cfef52efccdcfd31972f81c2e"

+[[package]]
+name = "byteview"
+version = "0.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6236364b88b9b6d0bc181ba374cf1ab55ba3ef97a1cb6f8cddad48a273767fb5"
+
 [[package]]
 name = "cc"
 version = "1.2.16"
@@ -798,6 +804,12 @@ version = "1.0.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "5b63caa9aa9397e2d9480a9b13673856c78d8ac123288526c37d7839f2a86990"

+[[package]]
+name = "compare"
+version = "0.0.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ea0095f6103c2a8b44acd6fd15960c801dafebf02e21940360833e0673f48ba7"
+
 [[package]]
 name = "core-foundation"
 version = "0.9.4"
@@ -838,6 +850,20 @@ version = "2.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "19d374276b40fb8bbdee95aef7c7fa6b5316ec764510eb64b8dd0e2ed0d7e7f5"

+[[package]]
+name = "crc-fast"
+version = "1.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bf0ab672ed7761a44410115d2e0e5ccd86b987ba186a78a4148fb02bb302a774"
+dependencies = [
+ "cc",
+ "crc",
+ "digest",
+ "libc",
+ "rand 0.9.1",
+ "regex",
+]
+
 [[package]]
 name = "crc32c"
 version = "0.6.8"
@@ -874,6 +900,15 @@ dependencies = [
 "crossbeam-utils",
 ]

+[[package]]
+name = "crossbeam-epoch"
+version = "0.9.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5b82ac4a3c2ca9c3460964f020e1402edd5753411d7737aa39c3714ad1b5420e"
+dependencies = [
+ "crossbeam-utils",
+]
+
 [[package]]
 name = "crossbeam-queue"
 version = "0.3.12"
@@ -883,6 +918,16 @@ dependencies = [
 "crossbeam-utils",
 ]

+[[package]]
+name = "crossbeam-skiplist"
+version = "0.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "df29de440c58ca2cc6e587ec3d22347551a32435fbde9d2bff64e78a9ffa151b"
+dependencies = [
+ "crossbeam-epoch",
+ "crossbeam-utils",
+]
+
 [[package]]
 name = "crossbeam-utils"
 version = "0.8.21"
@@ -896,7 +941,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1bfb12502f3fc46cca1bb51ac28df9d618d813cdc3d2f25b9fe775a34af26bb3"
 dependencies = [
 "generic-array",
- "rand_core",
+ "rand_core 0.6.4",
 "typenum",
 ]

@@ -954,6 +999,20 @@ dependencies = [
 "num_cpus",
 ]

+[[package]]
+name = "dashmap"
+version = "6.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5041cc499144891f3790297212f32a74fb938e5136a14943f338ef9e0ae276cf"
+dependencies = [
+ "cfg-if",
+ "crossbeam-utils",
+ "hashbrown 0.14.5",
+ "lock_api",
+ "once_cell",
+ "parking_lot_core 0.9.10",
+]
+
 [[package]]
 name = "deranged"
 version = "0.4.0"
@@ -996,6 +1055,12 @@ dependencies = [
 "syn 2.0.100",
 ]

+[[package]]
+name = "double-ended-peekable"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c0d05e1c0dbad51b52c38bda7adceef61b9efc2baf04acfe8726a8c4630a6f57"
+
 [[package]]
 name = "dyn-clone"
 version = "1.0.19"
@@ -1017,6 +1082,18 @@ dependencies = [
 "cfg-if",
 ]

+[[package]]
+name = "enum_dispatch"
+version = "0.3.13"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "aa18ce2bc66555b3218614519ac839ddb759a7d6720732f979ef8d13be147ecd"
+dependencies = [
+ "once_cell",
+ "proc-macro2",
+ "quote",
+ "syn 2.0.100",
+]
+
 [[package]]
 name = "env_logger"
 version = "0.10.2"
@@ -1084,6 +1161,23 @@ version = "0.4.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0ce7134b9999ecaf8bcd65542e436736ef32ddca1b3e06094cb6ec5755203b80"

+[[package]]
+name = "fjall"
+version = "2.11.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b25ad44cd4360a0448a9b5a0a6f1c7a621101cca4578706d43c9a821418aebc"
+dependencies = [
+ "byteorder",
+ "byteview",
+ "dashmap 6.1.0",
+ "log",
+ "lsm-tree",
+ "path-absolutize",
+ "std-semaphore",
+ "tempfile",
+ "xxhash-rust",
+]
+
 [[package]]
 name = "fnv"
 version = "1.0.7"
@@ -1200,7 +1294,7 @@ dependencies = [

 [[package]]
 name = "garage"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "assert-json-diff",
 "async-trait",
@@ -1211,7 +1305,7 @@ dependencies = [
 "bytes",
 "bytesize",
 "chrono",
- "crc32fast",
+ "crc-fast",
 "format_table",
 "futures",
 "garage_api_admin",
@@ -1249,13 +1343,14 @@ dependencies = [
 "timeago",
 "tokio",
 "tracing",
+ "tracing-journald",
 "tracing-subscriber",
 "utoipa",
 ]

 [[package]]
 name = "garage_api_admin"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "argon2",
 "async-trait",
@@ -1287,14 +1382,12 @@ dependencies = [

 [[package]]
 name = "garage_api_common"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "base64 0.21.7",
 "bytes",
 "chrono",
- "crc32c",
- "crc32fast",
- "crc64fast-nvme",
+ "crc-fast",
 "crypto-common",
 "err-derive",
 "futures",
@@ -1307,7 +1400,6 @@ dependencies = [
 "http-body-util",
 "hyper 1.6.0",
 "hyper-util",
- "idna 0.5.0",
 "md-5",
 "nom",
 "opentelemetry",
@@ -1323,7 +1415,7 @@ dependencies = [

 [[package]]
 name = "garage_api_k2v"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "base64 0.21.7",
 "err-derive",
@@ -1346,16 +1438,14 @@ dependencies = [

 [[package]]
 name = "garage_api_s3"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "aes-gcm",
 "async-compression",
 "base64 0.21.7",
 "bytes",
 "chrono",
- "crc32c",
- "crc32fast",
- "crc64fast-nvme",
+ "crc-fast",
 "err-derive",
 "form_urlencoded",
 "futures",
@@ -1393,7 +1483,7 @@ dependencies = [

 [[package]]
 name = "garage_block"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "arc-swap",
 "async-compression",
@@ -1407,7 +1497,7 @@ dependencies = [
 "garage_util",
 "hex",
 "opentelemetry",
- "rand",
+ "rand 0.8.5",
 "serde",
 "tokio",
 "tokio-util 0.7.14",
@@ -1417,11 +1507,13 @@ dependencies = [

 [[package]]
 name = "garage_db"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "err-derive",
+ "fjall",
 "heed",
 "mktemp",
+ "parking_lot 0.12.3",
 "r2d2",
 "r2d2_sqlite",
 "rusqlite",
@@ -1430,7 +1522,7 @@ dependencies = [

 [[package]]
 name = "garage_model"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "argon2",
 "async-trait",
@@ -1448,7 +1540,7 @@ dependencies = [
 "hex",
 "http 1.3.1",
 "parse_duration",
- "rand",
+ "rand 0.8.5",
 "serde",
 "serde_bytes",
 "tokio",
@@ -1458,7 +1550,7 @@ dependencies = [

 [[package]]
 name = "garage_net"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "arc-swap",
 "bytes",
@@ -1473,7 +1565,7 @@ dependencies = [
 "opentelemetry-contrib",
 "pin-project",
 "pretty_env_logger",
- "rand",
+ "rand 0.8.5",
 "rmp-serde",
 "serde",
 "tokio",
@@ -1483,7 +1575,7 @@ dependencies = [

 [[package]]
 name = "garage_rpc"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "arc-swap",
 "async-trait",
@@ -1503,7 +1595,7 @@ dependencies = [
 "nix",
 "opentelemetry",
 "pnet_datalink",
- "rand",
+ "rand 0.8.5",
 "reqwest",
 "schemars",
 "serde",
@@ -1515,7 +1607,7 @@ dependencies = [

 [[package]]
 name = "garage_table"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "arc-swap",
 "async-trait",
@@ -1527,7 +1619,7 @@ dependencies = [
 "hex",
 "hexdump",
 "opentelemetry",
- "rand",
+ "rand 0.8.5",
 "serde",
 "serde_bytes",
 "tokio",
@@ -1536,7 +1628,7 @@ dependencies = [

 [[package]]
 name = "garage_util"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "arc-swap",
 "async-trait",
@@ -1554,7 +1646,7 @@ dependencies = [
 "lazy_static",
 "mktemp",
 "opentelemetry",
- "rand",
+ "rand 0.8.5",
 "rmp-serde",
 "rustc_version",
 "serde",
@@ -1568,7 +1660,7 @@ dependencies = [

 [[package]]
 name = "garage_web"
-version = "2.0.0"
+version = "2.1.0"
 dependencies = [
 "err-derive",
 "garage_api_common",
@@ -1664,6 +1756,12 @@ dependencies = [
 "syn 2.0.100",
 ]

+[[package]]
+name = "guardian"
+version = "1.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "17e2ac29387b1aa07a1e448f7bb4f35b500787971e965b02842b900afa5c8f6f"
+
 [[package]]
 name = "h2"
 version = "0.3.26"
@@ -1731,11 +1829,11 @@ dependencies = [

 [[package]]
 name = "hashlink"
-version = "0.9.1"
+version = "0.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6ba4ff7128dee98c7dc9794b6a411377e1404dba1c97deb8d1a55297bd25d8af"
+checksum = "7382cf6263419f2d8df38c55d7da83da5c18aef87fc7a7fc1fb1e344edfe14c1"
 dependencies = [
- "hashbrown 0.14.5",
+ "hashbrown 0.15.2",
 ]

 [[package]]
@@ -2179,16 +2277,6 @@ version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b9e0384b61958566e926dc50660321d12159025e767c18e043daf26b70104c39"

-[[package]]
-name = "idna"
-version = "0.5.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "634d9b1461af396cad843f47fdba5597a4f9e6ddd4bfb6ff5d85028c25cb12f6"
-dependencies = [
- "unicode-bidi",
- "unicode-normalization",
-]
-
 [[package]]
 name = "idna"
 version = "1.0.3"
@@ -2249,6 +2337,15 @@ dependencies = [
 "cfg-if",
 ]

+[[package]]
+name = "interval-heap"
+version = "0.0.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "11274e5e8e89b8607cfedc2910b6626e998779b48a019151c7604d0adcb86ac6"
+dependencies = [
+ "compare",
+]
+
 [[package]]
 name = "ipnet"
 version = "2.11.0"
@@ -2542,9 +2639,9 @@ dependencies = [

 [[package]]
 name = "libsqlite3-sys"
-version = "0.28.0"
+version = "0.35.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0c10584274047cb335c23d3e61bcef8e323adae7c5c8c760540f73610177fc3f"
+checksum = "133c182a6a2c87864fe97778797e46c7e999672690dc9fa3ee8e241aa4a9c13f"
 dependencies = [
 "cc",
 "pkg-config",
@@ -2605,6 +2702,36 @@ dependencies = [
 "hashbrown 0.15.2",
 ]

+[[package]]
+name = "lsm-tree"
+version = "2.10.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ab73c02eadb3dc12c0024e5b61d6284e6d59064e67e74fbad77856caa56f62c7"
+dependencies = [
+ "byteorder",
+ "crossbeam-skiplist",
+ "double-ended-peekable",
+ "enum_dispatch",
+ "guardian",
+ "interval-heap",
+ "log",
+ "lz4_flex",
+ "path-absolutize",
+ "quick_cache",
+ "rustc-hash",
+ "self_cell",
+ "tempfile",
+ "value-log",
+ "varint-rs",
+ "xxhash-rust",
+]
+
+[[package]]
+name = "lz4_flex"
+version = "0.11.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "75761162ae2b0e580d7e7c390558127e5f01b4194debd6221fd8c207fc80e3f5"
+
 [[package]]
 name = "matchers"
 version = "0.1.0"
@@ -2859,7 +2986,7 @@ checksum = "6105e89802af13fdf48c49d7646d3b533a70e536d818aae7e78ba0433d01acb8"
 dependencies = [
 "async-trait",
 "crossbeam-channel",
- "dashmap",
+ "dashmap 4.0.2",
 "fnv",
 "futures-channel",
 "futures-executor",
@@ -2868,7 +2995,7 @@ dependencies = [
 "lazy_static",
 "percent-encoding",
 "pin-project",
- "rand",
+ "rand 0.8.5",
 "thiserror 1.0.69",
 "tokio",
 "tokio-stream",
@@ -3010,7 +3137,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "346f04948ba92c43e8469c1ee6736c7563d71012b17d40745260fe106aac2166"
 dependencies = [
 "base64ct",
- "rand_core",
+ "rand_core 0.6.4",
 "subtle",
 ]

@@ -3020,6 +3147,24 @@ version = "1.0.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "57c0d7b74b563b49d38dae00a0c37d4d6de9b432382b2892f0574ddcae73fd0a"

+[[package]]
+name = "path-absolutize"
+version = "3.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e4af381fe79fa195b4909485d99f73a80792331df0625188e707854f0b3383f5"
+dependencies = [
+ "path-dedot",
+]
+
+[[package]]
+name = "path-dedot"
+version = "3.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "07ba0ad7e047712414213ff67533e6dd477af0a4e1d14fb52343e53d30ea9397"
+dependencies = [
+ "once_cell",
+]
+
 [[package]]
 name = "pem"
 version = "3.0.5"
@@ -3315,6 +3460,16 @@ dependencies = [
 "serde",
 ]

+[[package]]
+name = "quick_cache"
+version = "0.6.16"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ad6644cb07b7f3488b9f3d2fde3b4c0a7fa367cafefb39dff93a659f76eb786"
+dependencies = [
+ "equivalent",
+ "hashbrown 0.15.2",
+]
+
 [[package]]
 name = "quote"
 version = "1.0.40"
@@ -3343,9 +3498,9 @@ dependencies = [

 [[package]]
 name = "r2d2_sqlite"
-version = "0.24.0"
+version = "0.31.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6a982edf65c129796dba72f8775b292ef482b40d035e827a9825b3bc07ccc5f2"
+checksum = "63417e83dc891797eea3ad379f52a5986da4bca0d6ef28baf4d14034dd111b0c"
 dependencies = [
 "r2d2",
 "rusqlite",
@@ -3359,8 +3514,18 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "34af8d1a0e25924bc5b7c43c079c942339d8f0a8b57c39049bef581b46327404"
 dependencies = [
 "libc",
- "rand_chacha",
- "rand_core",
+ "rand_chacha 0.3.1",
+ "rand_core 0.6.4",
+]
+
+[[package]]
+name = "rand"
+version = "0.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9fbfd9d094a40bf3ae768db9361049ace4c0e04a4fd6b359518bd7b73a73dd97"
+dependencies = [
+ "rand_chacha 0.9.0",
+ "rand_core 0.9.3",
 ]

 [[package]]
@@ -3370,7 +3535,17 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "e6c10a63a0fa32252be49d21e7709d4d4baf8d231c2dbce1eaa8141b9b127d88"
 dependencies = [
 "ppv-lite86",
- "rand_core",
+ "rand_core 0.6.4",
+]
+
+[[package]]
+name = "rand_chacha"
+version = "0.9.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d3022b5f1df60f26e1ffddd6c66e8aa15de382ae63b3a0c1bfc0e4d3e3f325cb"
+dependencies = [
+ "ppv-lite86",
+ "rand_core 0.9.3",
 ]

 [[package]]
@@ -3382,6 +3557,15 @@ dependencies = [
 "getrandom 0.2.15",
 ]

+[[package]]
+name = "rand_core"
+version = "0.9.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "99d9a13982dcf210057a8a78572b2217b667c3beacbf3a0d8b454f6f82837d38"
+dependencies = [
+ "getrandom 0.3.2",
+]
+
 [[package]]
 name = "redox_syscall"
 version = "0.2.16"
@@ -3534,9 +3718,9 @@ checksum = "3cd14fd5e3b777a7422cca79358c57a8f6e3a703d9ac187448d0daf220c2407f"

 [[package]]
 name = "rusqlite"
-version = "0.31.0"
+version = "0.37.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b838eba278d213a8beaf485bd313fd580ca4505a00d5871caeb1457c55322cae"
+checksum = "165ca6e57b20e1351573e3729b958bc62f0e48025386970b6e4d29e7a7e71f3f"
 dependencies = [
 "bitflags 2.9.0",
 "fallible-iterator",
@@ -3552,6 +3736,12 @@ version = "0.1.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "719b953e2095829ee67db738b3bfa9fa368c94900df327b3f07fe6e794d2fe1f"

+[[package]]
+name = "rustc-hash"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d"
+
 [[package]]
 name = "rustc_version"
 version = "0.4.1"
@@ -3794,6 +3984,12 @@ dependencies = [
 "libc",
 ]

+[[package]]
+name = "self_cell"
+version = "1.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0f7d95a54511e0c7be3f51e8867aa8cf35148d7b9445d44de2f943e2b206e749"
+
 [[package]]
 name = "semver"
 version = "1.0.26"
@@ -4007,6 +4203,12 @@ dependencies = [
 "syn 1.0.109",
 ]

+[[package]]
+name = "std-semaphore"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "33ae9eec00137a8eed469fb4148acd9fc6ac8c3f9b110f52cd34698c8b5bfa0e"
+
 [[package]]
 name = "strsim"
 version = "0.11.1"
@@ -4262,21 +4464,6 @@ dependencies = [
 "zerovec",
 ]

-[[package]]
-name = "tinyvec"
-version = "1.9.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09b3661f17e86524eccd4371ab0429194e0d7c008abb45f7a7495b1719463c71"
-dependencies = [
- "tinyvec_macros",
-]
-
-[[package]]
-name = "tinyvec_macros"
-version = "0.1.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
-
 [[package]]
 name = "tokio"
 version = "1.44.1"
@@ -4465,7 +4652,7 @@ dependencies = [
 "indexmap 1.9.3",
 "pin-project",
 "pin-project-lite",
- "rand",
+ "rand 0.8.5",
 "slab",
 "tokio",
 "tokio-util 0.7.14",
@@ -4550,6 +4737,17 @@ dependencies = [
 "tracing",
 ]

+[[package]]
+name = "tracing-journald"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fc0b4143302cf1022dac868d521e36e8b27691f72c84b3311750d5188ebba657"
+dependencies = [
+ "libc",
+ "tracing-core",
+ "tracing-subscriber",
+]
+
 [[package]]
 name = "tracing-log"
 version = "0.2.0"
@@ -4597,27 +4795,12 @@ version = "0.1.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2896d95c02a80c6d6a5d6e953d479f5ddf2dfdb6a244441010e373ac0fb88971"

-[[package]]
-name = "unicode-bidi"
-version = "0.3.18"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5c1cb5db39152898a79168971543b1cb5020dff7fe43c8dc468b0885f5e29df5"
-
 [[package]]
 name = "unicode-ident"
 version = "1.0.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "5a5f39404a5da50712a4c1eecf25e90dd62b613502b7e925fd4e4d19b5c96512"

-[[package]]
-name = "unicode-normalization"
-version = "0.1.24"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5033c97c4262335cded6d6fc3e5c18ab755e1a3dc96376350f3d8e9f009ad956"
-dependencies = [
- "tinyvec",
-]
-
 [[package]]
 name = "unicode-segmentation"
 version = "1.12.0"
@@ -4665,7 +4848,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "32f8b686cadd1473f4bd0117a5d28d36b1ade384ea9b5069a1c40aefed7fda60"
 dependencies = [
 "form_urlencoded",
- "idna 1.0.3",
+ "idna",
 "percent-encoding",
 ]

@@ -4717,7 +4900,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "79daa5ed5740825c40b389c5e50312b9c86df53fccd33f281df655642b43869d"
 dependencies = [
 "getrandom 0.2.15",
- "rand",
+ "rand 0.8.5",
 ]

 [[package]]
@@ -4726,6 +4909,29 @@ version = "0.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ba73ea9cf16a25df0c8caa16c51acb937d5712a8429db78a3ee29d5dcacd3a65"

+[[package]]
+name = "value-log"
+version = "1.9.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "62fc7c4ce161f049607ecea654dca3f2d727da5371ae85e2e4f14ce2b98ed67c"
+dependencies = [
+ "byteorder",
+ "byteview",
+ "interval-heap",
+ "log",
+ "path-absolutize",
+ "rustc-hash",
+ "tempfile",
+ "varint-rs",
+ "xxhash-rust",
+]
+
+[[package]]
+name = "varint-rs"
+version = "2.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8f54a172d0620933a27a4360d3db3e2ae0dd6cceae9730751a036bbf182c4b23"
+
 [[package]]
 name = "vcpkg"
 version = "0.2.15"
@@ -24,18 +24,18 @@ default-members = ["src/garage"]

 # Internal Garage crates
 format_table = { version = "0.1.1", path = "src/format-table" }
-garage_api_common = { version = "2.0.0", path = "src/api/common" }
-garage_api_admin = { version = "2.0.0", path = "src/api/admin" }
-garage_api_s3 = { version = "2.0.0", path = "src/api/s3" }
-garage_api_k2v = { version = "2.0.0", path = "src/api/k2v" }
-garage_block = { version = "2.0.0", path = "src/block" }
-garage_db = { version = "2.0.0", path = "src/db", default-features = false }
-garage_model = { version = "2.0.0", path = "src/model", default-features = false }
-garage_net = { version = "2.0.0", path = "src/net" }
-garage_rpc = { version = "2.0.0", path = "src/rpc" }
-garage_table = { version = "2.0.0", path = "src/table" }
-garage_util = { version = "2.0.0", path = "src/util" }
-garage_web = { version = "2.0.0", path = "src/web" }
+garage_api_common = { version = "2.1.0", path = "src/api/common" }
+garage_api_admin = { version = "2.1.0", path = "src/api/admin" }
+garage_api_s3 = { version = "2.1.0", path = "src/api/s3" }
+garage_api_k2v = { version = "2.1.0", path = "src/api/k2v" }
+garage_block = { version = "2.1.0", path = "src/block" }
+garage_db = { version = "2.1.0", path = "src/db", default-features = false }
+garage_model = { version = "2.1.0", path = "src/model", default-features = false }
+garage_net = { version = "2.1.0", path = "src/net" }
+garage_rpc = { version = "2.1.0", path = "src/rpc" }
+garage_table = { version = "2.1.0", path = "src/table" }
+garage_util = { version = "2.1.0", path = "src/util" }
+garage_web = { version = "2.1.0", path = "src/web" }
 k2v-client = { version = "0.0.4", path = "src/k2v-client" }

 # External crates from crates.io
@@ -49,9 +49,7 @@ bytes = "1.0"
 bytesize = "1.1"
 cfg-if = "1.0"
 chrono = { version = "0.4", features = ["serde"] }
-crc32fast = "1.4"
-crc32c = "0.6"
-crc64fast-nvme = "1.2"
+crc-fast = "1.1"
 crypto-common = "0.1"
 err-derive = "0.3"
 gethostname = "0.4"
@@ -59,7 +57,6 @@ git-version = "0.3.4"
 hex = "0.4"
 hexdump = "0.1"
 hmac = "0.12"
-idna = "0.5"
 itertools = "0.12"
 ipnet = "2.9.0"
 lazy_static = "1.4"
@@ -67,6 +64,7 @@ md-5 = "0.10"
 mktemp = "0.5"
 nix = { version = "0.29", default-features = false, features = ["fs"] }
 nom = "7.1"
+parking_lot = "0.12"
 parse_duration = "2.1"
 paste = "1.0"
 pin-project = "1.0.12"
@@ -86,12 +84,14 @@ pretty_env_logger = "0.5"
 structopt = { version = "0.3", default-features = false }
 syslog-tracing = "0.3"
 tracing = "0.1"
+tracing-journald = "0.3.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }

 heed = { version = "0.11", default-features = false, features = ["lmdb"] }
-rusqlite = "0.31.0"
+rusqlite = "0.37"
 r2d2 = "0.8"
-r2d2_sqlite = "0.24"
+r2d2_sqlite = "0.31"
+fjall = "2.4"

 async-compression = { version = "0.4", features = ["tokio", "zstd"] }
 zstd = { version = "0.13", default-features = false }
@@ -1,7 +1,7 @@
 <!DOCTYPE html>
 <html>
  <head>
-    <title>Garage adminstration API v0</title>
+    <title>Garage administration API v0</title>
    <!-- needed for adaptive design -->
    <meta charset="utf-8"/>
    <meta name="viewport" content="width=device-width, initial-scale=1">
@@ -1,7 +1,7 @@
 <!DOCTYPE html>
 <html>
  <head>
-    <title>Garage adminstration API v1</title>
+    <title>Garage administration API v1</title>
    <!-- needed for adaptive design -->
    <meta charset="utf-8"/>
    <meta name="viewport" content="width=device-width, initial-scale=1">
@@ -1,7 +1,7 @@
 <!DOCTYPE html>
 <html>
  <head>
-    <title>Garage adminstration API v2</title>
+    <title>Garage administration API v2</title>
    <!-- needed for adaptive design -->
    <meta charset="utf-8"/>
    <meta name="viewport" content="width=device-width, initial-scale=1">
@@ -12,7 +12,7 @@
      "name": "AGPL-3.0",
      "identifier": "AGPL-3.0"
    },
-    "version": "v2.0.0"
+    "version": "v2.1.0"
  },
  "servers": [
    {
@@ -31,9 +31,12 @@
        "parameters": [
          {
            "name": "domain",
-            "in": "path",
+            "in": "query",
            "description": "The domain name to check for",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -405,9 +408,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -437,9 +443,12 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Admin API token ID",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -462,9 +471,12 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "ID of the bucket to delete",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -493,9 +505,12 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Access key ID",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -552,26 +567,20 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Admin API token ID",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          },
          {
            "name": "search",
-            "in": "path",
+            "in": "query",
            "description": "Partial token ID or name to search for",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          }
        ],
@@ -602,9 +611,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -644,38 +656,29 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Exact bucket ID to look up",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          },
          {
            "name": "globalAlias",
-            "in": "path",
+            "in": "query",
            "description": "Global alias of bucket to look up",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          },
          {
            "name": "search",
-            "in": "path",
+            "in": "query",
            "description": "Partial ID or alias to search for",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          }
        ],
@@ -813,6 +816,30 @@
        }
      }
    },
+    "/v2/GetCurrentAdminTokenInfo": {
+      "get": {
+        "tags": [
+          "Admin API token"
+        ],
+        "description": "\nReturn information about the calling admin API token.\n    ",
+        "operationId": "GetCurrentAdminTokenInfo",
+        "responses": {
+          "200": {
+            "description": "Information about the admin token",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/GetCurrentAdminTokenInfoResponse"
+                }
+              }
+            }
+          },
+          "500": {
+            "description": "Internal server error"
+          }
+        }
+      }
+    },
    "/v2/GetKeyInfo": {
      "get": {
        "tags": [
@@ -823,33 +850,27 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Access key ID",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          },
          {
            "name": "search",
-            "in": "path",
+            "in": "query",
            "description": "Partial key ID or name to search for",
-            "required": true,
+            "required": false,
            "schema": {
-              "type": [
-                "string",
-                "null"
-              ]
+              "type": "string"
            }
          },
          {
            "name": "showSecretKey",
-            "in": "path",
+            "in": "query",
            "description": "Whether to return the secret access key",
-            "required": true,
+            "required": false,
            "schema": {
              "type": "boolean"
            }
@@ -882,9 +903,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -914,9 +938,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -946,9 +973,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -988,9 +1018,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1064,7 +1097,7 @@
        "parameters": [
          {
            "name": "bucketId",
-            "in": "path",
+            "in": "query",
            "required": true,
            "schema": {
              "type": "string"
@@ -1072,7 +1105,7 @@
          },
          {
            "name": "key",
-            "in": "path",
+            "in": "query",
            "required": true,
            "schema": {
              "type": "string"
@@ -1109,9 +1142,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1175,9 +1211,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "responses": {
@@ -1255,9 +1294,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1321,9 +1363,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1397,9 +1442,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1463,9 +1511,12 @@
        "parameters": [
          {
            "name": "node",
-            "in": "path",
+            "in": "query",
            "description": "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1505,9 +1556,12 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Admin API token ID",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1547,9 +1601,12 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "ID of the bucket to update",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -1627,9 +1684,12 @@
        "parameters": [
          {
            "name": "id",
-            "in": "path",
+            "in": "query",
            "description": "Access key ID",
-            "required": true
+            "required": true,
+            "schema": {
+              "type": "string"
+            }
          }
        ],
        "requestBody": {
@@ -2383,7 +2443,7 @@
          "knownNodes",
          "connectedNodes",
          "storageNodes",
-          "storageNodesOk",
+          "storageNodesUp",
          "partitions",
          "partitionsQuorum",
          "partitionsAllOk"
@@ -2423,7 +2483,7 @@
            "description": "the number of storage nodes currently registered in the cluster layout",
            "minimum": 0
          },
-          "storageNodesOk": {
+          "storageNodesUp": {
            "type": "integer",
            "description": "the number of storage nodes to which a connection is currently open",
            "minimum": 0
@@ -2558,6 +2618,9 @@
          }
        }
      },
+      "GetCurrentAdminTokenInfoResponse": {
+        "$ref": "#/components/schemas/GetAdminTokenInfoResponse"
+      },
      "GetKeyInfoResponse": {
        "type": "object",
        "required": [
@@ -3120,9 +3183,15 @@
          "blocksPurged",
          "objectsDeleted",
          "uploadsDeleted",
-          "versionsDeleted"
+          "versionsDeleted",
+          "blockRefsPurged"
        ],
        "properties": {
+          "blockRefsPurged": {
+            "type": "integer",
+            "format": "int64",
+            "minimum": 0
+          },
          "blocksPurged": {
            "type": "integer",
            "format": "int64",
@@ -3574,9 +3643,15 @@
                "blocksPurged",
                "objectsDeleted",
                "uploadsDeleted",
-                "versionsDeleted"
+                "versionsDeleted",
+                "blockRefsPurged"
              ],
              "properties": {
+                "blockRefsPurged": {
+                  "type": "integer",
+                  "format": "int64",
+                  "minimum": 0
+                },
                "blocksPurged": {
                  "type": "integer",
                  "format": "int64",
@@ -3973,6 +4048,18 @@
                "$ref": "#/components/schemas/ScrubCommand"
              }
            }
+          },
+          {
+            "type": "string",
+            "enum": [
+              "aliases"
+            ]
+          },
+          {
+            "type": "string",
+            "enum": [
+              "clearResyncQueue"
+            ]
          }
        ]
      },
@@ -12,7 +12,7 @@ In this section, we cover the following web applications:
 | [Mastodon](#mastodon)     | ✅       | Natively supported    |
 | [Matrix](#matrix)     | ✅       |  Tested with `synapse-s3-storage-provider`    |
 | [ejabberd](#ejabberd)     | ✅       |  `mod_s3_upload`    |
-| [Pixelfed](#pixelfed)     | ❓       |  Not yet tested    |
+| [Pixelfed](#pixelfed)     | ✅       |  Natively supported    |
 | [Pleroma](#pleroma)     | ❓       |  Not yet tested    |
 | [Lemmy](#lemmy)     | ✅        |  Supported with pict-rs    |
 | [Funkwhale](#funkwhale)     | ❓       | Not yet tested     |
@@ -69,7 +69,7 @@ $CONFIG = array(
        'hostname' => '127.0.0.1', // Can also be a domain name, eg. garage.example.com
        'port' => 3900,            // Put your reverse proxy port or your S3 API port
        'use_ssl' => false,        // Set it to true if you have a TLS enabled reverse proxy
-        'region' => 'garage',      // Garage has only one region named "garage"
+        'region' => 'garage',      // Garage default region is named "garage", edit according to your cluster config
        'use_path_style' => true   // Garage supports only path style, must be set to true
    ],
 ],
@@ -135,7 +135,7 @@ bucket but doesn't also know the secret encryption key.
 *Click on the picture to zoom*

 Add a new external storage. Put what you want in "folder name" (eg. "shared"). Select "Amazon S3". Keep "Access Key" for the Authentication field.
-In Configuration, put your bucket name (eg. nextcloud), the host (eg. 127.0.0.1), the port (eg. 3900 or 443), the region (garage). Tick the SSL box if you have put an HTTPS proxy in front of garage. You must tick the "Path access" box and you must leave the "Legacy authentication (v2)" box empty. Put your Key ID (eg. GK...) and your Secret Key in the last two input boxes. Finally click on the tick symbol on the right of your screen.
+In Configuration, put your bucket name (eg. nextcloud), the host (eg. 127.0.0.1), the port (eg. 3900 or 443), the region ("garage" if you use the default, or the one your configured in your `garage.toml`). Tick the SSL box if you have put an HTTPS proxy in front of garage. You must tick the "Path access" box and you must leave the "Legacy authentication (v2)" box empty. Put your Key ID (eg. GK...) and your Secret Key in the last two input boxes. Finally click on the tick symbol on the right of your screen.

 Now go to your "Files" app and a new "linked folder" has appeared with the name you chose earlier (eg. "shared").

@@ -191,10 +191,10 @@ garage key create peertube-key

 Keep the Key ID and the Secret key in a pad, they will be needed later.  

-We need two buckets, one for normal videos (named peertube-video) and one for webtorrent videos (named peertube-playlist).
+We need two buckets, one for normal videos (named peertube-videos) and one for webtorrent videos (named peertube-playlists).
 ```bash
 garage bucket create peertube-videos
-garage bucket create peertube-playlist
+garage bucket create peertube-playlists
 ```

 Now we allow our key to read and write on these buckets:
@@ -238,7 +238,7 @@ object_storage:
  # Put localhost only if you have a garage instance running on that node
  endpoint: 'http://localhost:3900' # or "garage.example.com" if you have TLS on port 443

-  # Garage supports only one region for now, named garage
+  # Garage default region is named "garage", edit according to your config
  region: 'garage'

  credentials:
@@ -253,7 +253,7 @@ object_storage:
    proxify_private_files: false

  streaming_playlists:
-    bucket_name: 'peertube-playlist'
+    bucket_name: 'peertube-playlists'

    # Keep it empty for our example
    prefix: ''
@@ -441,7 +441,7 @@ media_storage_providers:
  store_synchronous: True  # do we want to wait that the file has been written before returning?
  config:
    bucket: matrix       # the name of our bucket, we chose matrix earlier
-    region_name: garage  # only "garage" is supported for the region field
+    region_name: garage  # "garage" by default, edit according to your cluster config
    endpoint_url: http://localhost:3900 # the path to the S3 endpoint
    access_key_id: "GKxxx" # your Key ID
    secret_access_key: "xxxx" # your Secret Key
@@ -161,3 +161,49 @@ kopia repository validate-provider

 You can then run all the standard kopia commands: `kopia snapshot create`, `kopia mount`...
 Everything should work out-of-the-box.
+
+## Plakar
+
+Create your key and bucket on Garage server:
+
+```bash
+garage key create my-plakar-key
+garage bucket create plakar-backups
+garage bucket allow plakar-backups --read --write --key my-plakar-key
+…
+```
+
+On Plakar server, add your Garage as a storage location:
+```bash
+plakar store add garageS3 s3://my-garage.tld/plakar-backups \
+region=garage # Or as you've specified in garage.toml \
+access_key=<Key ID from "garage key info my-plakar-key"> \
+secret_access_key=<Secret key from "garage key info my-plakar-key">
+```
+
+Then create the repository.
+```bash
+plakar at @garageS3 create -plaintext # Unencrypted
+# or
+plakar at @garageS3 create #encrypted
+```
+
+If you encrypt your backups (Plakar default), you will need to define a strong passphrase. Do not forget to save your password safely. It will be needed to decrypt your backups.
+
+
+After the repository has been created, check that everything works as expected (that might give an empty result as no file has been added yet, but no error message):
+```bash
+plakar at @garageS3 check
+```
+
+Now that everything is configure, you can use Garage as your backups storage. For instance sync it with a local backup storage:
+```bash
+$ plakar at ~/backups sync to @garageS3
+```
+
+Or list the S3 storage content:
+```bash
+$ plakar at @garageS3 ls
+```
+
+More information in Plakar documentation: https://www.plakar.io/docs/main/quickstart/
@@ -8,18 +8,18 @@ have published Ansible roles.  We list them and compare them below.

 ## Comparison of Ansible roles

-| Feature                            | [ansible-role-garage](#zorun-ansible-role-garage) | [garage-docker-ansible-deploy](#moan0s-garage-docker-ansible-deploy) |
-|------------------------------------|---------------------------------------------|---------------------------------------------------------------|
-| **Runtime**                        | Systemd                                     | Docker                                                        |
-| **Target OS**                      | Any Linux                                   | Any Linux                                                     |
-| **Architecture**                   | amd64, arm64, i686                          | amd64, arm64                                                  |
-| **Additional software**            | None                                        | Traefik                                                       |
-| **Automatic node connection**      | ❌                                          | ✅                                                            |
-| **Layout management**              | ❌                                          | ✅                                                            |
-| **Manage buckets & keys**          | ❌                                          | ✅ (basic)                                                    |
-| **Allow custom Garage config**     | ✅                                          | ❌                                                            |
-| **Facilitate Garage upgrades**     | ✅                                          | ❌                                                            |
-| **Multiple instances on one host** | ✅                                          | ✅                                                            |
+| Feature                            | [ansible-role-garage](#zorun-ansible-role-garage) | [garage-docker-ansible-deploy](#moan0s-garage-docker-ansible-deploy) | [eddster2309 ansible-role-garage](#eddster2309-ansible-role-garage) |
+|------------------------------------|---------------------------------------------|---------------------------------------------------------------|---------------------------------|
+| **Runtime**                        | Systemd                                     | Docker                                                        | Systemd                         |
+| **Target OS**                      | Any Linux                                   | Any Linux                                                     | Any Linux                       |
+| **Architecture**                   | amd64, arm64, i686                          | amd64, arm64                                                  | arm64, arm, 386, amd64                            |
+| **Additional software**            | None                                        | Traefik                                                       | Ngnix and Keepalived (optional) |
+| **Automatic node connection**      | ❌                                          | ✅                                                            | ✅                              |
+| **Layout management**              | ❌                                          | ✅                                                            | ✅                              |
+| **Manage buckets & keys**          | ❌                                          | ✅ (basic)                                                    | ✅                              |
+| **Allow custom Garage config**     | ✅                                          | ❌                                                            | ❌                              |
+| **Facilitate Garage upgrades**     | ✅                                          | ❌                                                            | ✅                              |
+| **Multiple instances on one host** | ✅                                          | ✅                                                            | ❌                              |


 ## zorun/ansible-role-garage
@@ -49,3 +49,15 @@ structured DNS names, etc).

 As a result, this role makes it easier to start with Garage on Ansible,
 but is less flexible.
+
+## eddster2309/ansible-role-garage
+
+[Source code](https://github.com/eddster2309/ansible-role-garage), [Ansible galaxy](https://galaxy.ansible.com/ui/standalone/roles/eddster2309/garage/)
+
+This role is a opinionated but customisable role using the official Garage
+static binaries and only requires Systemd. As such it should work on any
+Linux based host. It includes all the nesscary configuration to
+automatically setup a clustered Garage deployment. Most Garage
+configuration options are exposed through Ansible variables so while you
+can't provide a custom config you can get very close. It can optionally
+installed a HA nginx deployment with Keepalived.
@@ -26,6 +26,13 @@ Or deploy with custom values:
 helm install --create-namespace --namespace garage garage ./garage -f values.override.yaml
 ```

+If you want to manage the CustomRessourceDefinition used by garage for its `kubernetes_discovery` outside of the helm chart, add `garage.kubernetesSkipCrd: true` to your custom values and use the kustomization before deploying the helm chart:
+
+```bash
+kubectl apply -k ../k8s/crd
+helm install --create-namespace --namespace garage garage ./garage -f values.override.yaml
+```
+
 After deploying, cluster layout must be configured manually as described in [Creating a cluster layout](@/documentation/quick-start/_index.md#creating-a-cluster-layout). Use the following command to access garage CLI:

 ```bash
@@ -45,7 +52,7 @@ This is an example `values.overrride.yaml` for deploying in a microk8s cluster w
 ```yaml
 garage:
  # Use only 2 replicas per object
-  replicationMode: "2"
+  replicationFactor: 2

 # Start 4 instances (StatefulSets) of garage
 deployment:
@@ -96,14 +96,14 @@ to store 2 TB of data in total.
 ## Get a Docker image

 Our docker image is currently named `dxflrs/garage` and is stored on the [Docker Hub](https://hub.docker.com/r/dxflrs/garage/tags?page=1&ordering=last_updated).
-We encourage you to use a fixed tag (eg. `v2.0.0`) and not the `latest` tag.
-For this example, we will use the latest published version at the time of the writing which is `v2.0.0` but it's up to you
+We encourage you to use a fixed tag (eg. `v2.1.0`) and not the `latest` tag.
+For this example, we will use the latest published version at the time of the writing which is `v2.1.0` but it's up to you
 to check [the most recent versions on the Docker Hub](https://hub.docker.com/r/dxflrs/garage/tags?page=1&ordering=last_updated).

 For example:

 ```
-sudo docker pull dxflrs/garage:v2.0.0
+sudo docker pull dxflrs/garage:v2.1.0
 ```

 ## Deploying and configuring Garage
@@ -171,7 +171,7 @@ docker run \
  -v /etc/garage.toml:/etc/garage.toml \
  -v /var/lib/garage/meta:/var/lib/garage/meta \
  -v /var/lib/garage/data:/var/lib/garage/data \
-  dxflrs/garage:v2.0.0
+  dxflrs/garage:v2.1.0
 ```

 With this command line, Garage should be started automatically at each boot.
@@ -185,7 +185,7 @@ If you want to use `docker-compose`, you may use the following `docker-compose.y
 version: "3"
 services:
  garage:
-    image: dxflrs/garage:v2.0.0
+    image: dxflrs/garage:v2.1.0
    network_mode: "host"
    restart: unless-stopped
    volumes:
@@ -97,7 +97,7 @@ server {
  location / {
    proxy_pass http://s3_backend;
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-    proxy_set_header Host $host;
+    proxy_set_header Host $http_host;
    # Disable buffering to a temporary file.
    proxy_max_temp_file_size 0;
  }
@@ -28,6 +28,7 @@ StateDirectory=garage
 DynamicUser=true
 ProtectHome=true
 NoNewPrivileges=true
+LimitNOFILE=42000

 [Install]
 WantedBy=multi-user.target
@@ -59,11 +59,13 @@ Garage themselves for the following tasks:

 - Hosting of their homepage, [privacyguides.org](https://www.privacyguides.org/), and various other static sites

- As a Mastodon object storage backend for [mstdn.party](https://mstdn.party/) and [mstdn.plus](https://mstdn.plus/)
+- As a PowerDNS authoritative zone backend through [Lightning Stream](https://doc.powerdns.com/lightningstream/latest/index.html) and [LMDB](https://doc.powerdns.com/authoritative/backends/lmdb.html)
+
+- As a Mastodon media storage backend for [mstdn.party](https://mstdn.party/) and [mstdn.plus](https://mstdn.plus/)

 - As a PeerTube storage backend for [neat.tube](https://neat.tube/)

 - As a [Matrix media backend](https://github.com/matrix-org/synapse-s3-storage-provider)

 Triplebit's Garage cluster is a multi-site cluster currently composed of
-10 nodes in 3 physical locations.
+15 storage nodes in 3 physical locations.
@@ -42,7 +42,7 @@ You may pause an ongoing scrub using `garage repair scrub pause`, but note that
 the scrub will resume automatically 24 hours later as Garage will not let your
 cluster run without a regular scrub. If the scrub procedure is too intensive
 for your servers and is slowing down your workload, the recommended solution
-is to increase the "scrub tranquility" using `garage repair scrub set-tranquility`.
+is to increase the "scrub tranquility" using `garage worker set scrub-tranquility`.
 A higher tranquility value will make Garage take longer pauses between two block
 verifications. Of course, scrubbing the entire data store will also take longer.

@@ -129,10 +129,10 @@ docker run \
  -d \
  --name garaged \
  -p 3900:3900 -p 3901:3901 -p 3902:3902 -p 3903:3903 \
-  -v /etc/garage.toml:/path/to/garage.toml \
-  -v /var/lib/garage/meta:/path/to/garage/meta \
-  -v /var/lib/garage/data:/path/to/garage/data \
-  dxflrs/garage:v2.0.0
+  -v /path/to/garage.toml:/etc/garage.toml \
+  -v /path/to/garage/meta:/var/lib/garage/meta \
+  -v /path/to/garage/data:/var/lib/garage/data \
+  dxflrs/garage:v2.1.0
 ```

 Under Linux, you can substitute `--network host` for `-p 3900:3900 -p 3901:3901 -p 3902:3902 -p 3903:3903`
@@ -182,11 +182,12 @@ ID                 Hostname  Address         Tag                   Zone  Capacit
 ## Creating a cluster layout

 Creating a cluster layout for a Garage deployment means informing Garage
-of the disk space available on each node of the cluster
-as well as the zone (e.g. datacenter) each machine is located in.
+of the disk space available on each node of the cluster, `-c`,
+as well as the name of the zone (e.g. datacenter), `-z`, each machine is located in.

-For our test deployment, we are using only one node. The way in which we configure
-it does not matter, you can simply write:
+For our test deployment, we are have only one node with zone named `dc1` and a
+capacity of `1G`, though the capacity is ignored for a single node deployment
+and can be changed later when adding new nodes.

 ```bash
 garage layout assign -z dc1 -c 1G <node_id>
@@ -6,41 +6,167 @@ weight = 40
 The Garage administration API is accessible through a dedicated server whose
 listen address is specified in the `[admin]` section of the configuration
 file (see [configuration file
-reference](@/documentation/reference-manual/configuration.md))
+reference](@/documentation/reference-manual/configuration.md)).

-**WARNING.** At this point, there is no commitment to the stability of the APIs described in this document.
-We will bump the version numbers prefixed to each API endpoint each time the syntax
-or semantics change, meaning that code that relies on these endpoint will break
-when changes are introduced.
-
-Versions:
- - Before Garage 0.7.2 - no admin API
- - Garage 0.7.2 - admin APIv0
- - Garage 0.9.0 - admin APIv1, deprecate admin APIv0
+The current version of the admin API is v2.  No breaking changes to the Garage
+administration API will be published outside of a major release.

+History of previous versions:

+ - Before Garage v0.7.2 - no admin API
+ - Garage v0.7.2 - admin API v0
+ - Garage v0.9.0 - admin API v1, deprecate admin API v0
+ - Garage v2.0.0 - admin API v2, deprecate admin API v1

 ## Access control

-The admin API uses two different tokens for access control, that are specified in the config file's `[admin]` section:
+### Using an API token

- `metrics_token`: the token for accessing the Metrics endpoint (if this token
-  is not set in the config file, the Metrics endpoint can be accessed without
-  access control);
-
- `admin_token`: the token for accessing all of the other administration
-  endpoints (if this token is not set in the config file, access to these
-  endpoints is disabled entirely).
-
-These tokens are used as simple HTTP bearer tokens. In other words, to
-authenticate access to an admin API endpoint, add the following HTTP header
-to your request:
+Administration API tokens tokens are used as simple HTTP bearer tokens. In
+other words, to authenticate access to an admin API endpoint, add the following
+HTTP header to your request:

 ```
 Authorization: Bearer <token>
 ```

-## Administration API endpoints
+### User-defined API tokens
+
+Cluster administrators may dynamically define administration tokens using the CLI commands under `garage admin-token`.
+Such tokens may be limited in scope, meaning that they may enable access to only a subset of API calls.
+They may also have an expiration date to limit their use in time.
+
+Here is an example to create an administration token that is valid for 30 days
+and gives access to only a subset of API calls, allowing it to create buckets
+and access keys and give keys permissions on buckets:
+
+```bash
+$ garage admin-token create --expires-in 30d \
+    --scope ListBuckets,GetBucketInfo,ListKeys,GetKeyInfo,CreateBucket,CreateKey,AllowBucketKey,DenyBucketKey \
+    my-token
+This is your secret bearer token, it will not be shown again by Garage:
+
+  8ed1830b10a276ff57061950.kOSIpxWK9zSGbTO9Xadpv3YndSFWma0_snXcYHaORXk
+
+==== ADMINISTRATION TOKEN INFORMATION ====
+Token ID:    8ed1830b10a276ff57061950
+Token name:  my-token
+Created:     2025-06-15 15:12:44.160 +02:00
+Validity:    valid
+Expiration:  2025-07-15 15:12:44.117 +02:00
+
+Scope:       ListBuckets
+             GetBucketInfo
+             ListKeys
+             GetKeyInfo
+             CreateBucket
+             CreateKey
+             AllowBucketKey
+             DenyBucketKey
+```
+
+When running this command, your token will be shown only once and **will never
+be shown again by Garage**, so make sure to save it directly.  The token is
+hashed internally, and is identified by its prefix (32 hex digits followed by a
+dot) which is saved in clear.
+
+When running `garage admin-token list`, you might see something like this:
+
+```
+ID                        Created     Name                                       Expiration                      Scope
+-                         -           metrics_token (from daemon configuration)  never                           Metrics
+8ed1830b10a276ff57061950  2025-06-15  my-token                                   2025-07-15 15:12:44.117 +02:00  ListBuckets, ... (8)
+```
+
+### Master API tokens
+
+The admin API can also use two different master tokens for access control,
+specified in the config file's `[admin]` section:
+
+- `metrics_token`: the token for accessing the Metrics endpoint. If this token
+  is not set in the config file, the Metrics endpoint can be accessed without
+  access control.
+
+- `admin_token`: the token for accessing all of the other administration
+  endpoints. If this token is not set in the config file, access to these
+  endpoints is only possible with a user-defined admin token.
+
+With the introduction of multiple user-defined admin tokens, the use of master
+API tokens is now discouraged.
+
+
+## Using the admin API
+
+All of the admin API endpoints are described in the OpenAPI specification:
+
+ - APIv2 - [HTML spec](https://garagehq.deuxfleurs.fr/api/garage-admin-v2.html) - [OpenAPI JSON](https://garagehq.deuxfleurs.fr/api/garage-admin-v2.json)
+ - APIv1 (deprecated) - [HTML spec](https://garagehq.deuxfleurs.fr/api/garage-admin-v1.html) - [OpenAPI YAML](https://garagehq.deuxfleurs.fr/api/garage-admin-v1.yml)
+ - APIv0 (deprecated) - [HTML spec](https://garagehq.deuxfleurs.fr/api/garage-admin-v0.html) - [OpenAPI YAML](https://garagehq.deuxfleurs.fr/api/garage-admin-v0.yml)
+
+Making a request to the API from the command line can be as simple as running:
+
+```bash
+curl -H 'Authorization: Bearer s3cr3t' http://localhost:3903/v2/GetClusterStatus | jq
+```
+
+For more advanced use cases, we recommend using an SDK.  
+[Go to the "Build your own app" section to know how to use our SDKs](@/documentation/build/_index.md)
+
+### Making API calls from the `garage` CLI
+
+Since v2.0.0, the `garage` binary provides a subcommand `garage json-api` that
+allows you to invoke the API without making an HTTP request.  This can be
+useful for scripting Garage deployments.
+
+`garage json-api` proxies API calls through Garage's internal RPC protocol,
+therefore it does not require any form of authentication: RPC connection
+parameters are discovered automatically to contact the locally-running Garage
+instance (as when running any other `garage` CLI command).
+
+For simple calls that take no parameters, usage is as follows:
+
+```
+$ garage json-api GetClusterHealth
+{
+  "connectedNodes": 3,
+  "knownNodes": 3,
+  "partitions": 256,
+  "partitionsAllOk": 256,
+  "partitionsQuorum": 256,
+  "status": "healthy",
+  "storageNodes": 3,
+  "storageNodesOk": 3
+}
+```
+
+If you need to specify a JSON body for your call, you can add it directly after
+the name of the function you are calling:
+
+```
+$ garage json-api CreateAdminToken '{"name": "test"}'
+```
+
+Or you can feed it through stdin by adding a `-` as the last command parameter:
+
+```
+$ garage json-api CreateAdminToken -
+{"name": "test"}
+<EOF>
+```
+
+For admin API calls that would have taken query parameters in their HTTP version, these parameters can be passed in the JSON body object:
+
+```
+$ garage json-api GetAdminTokenInfo '{"id":"b0e6e0ace2c0b2aca4cdb2de"}'
+```
+
+For admin API calls that take both query parameters and a JSON body, combine them in the following fashion:
+
+```
+$ garage json-api UpdateAdminToken '{"id":"b0e6e0ace2c0b2aca4cdb2de", "body":{"name":"not a test"}}'
+```
+
+## Special administration API endpoints

 ### Metrics `GET /metrics`

@@ -83,7 +209,7 @@ content-length: 102
 date: Tue, 08 Aug 2023 07:22:38 GMT

 Garage is fully operational
-Consult the full health check API endpoint at /v0/health for more details
+Consult the full health check API endpoint at /v2/GetClusterHealth for more details
 ```

 ### On-demand TLS `GET /check`
@@ -126,23 +252,7 @@ $ curl -so /dev/null -w "%{http_code}" http://localhost:3903/check?domain=exampl
 200
 ```

-
 **References:**
 - [Using On-Demand TLS](https://caddyserver.com/docs/automatic-https#using-on-demand-tls)
 - [Add option for a backend check to approve use of on-demand TLS](https://github.com/caddyserver/caddy/pull/1939)
 - [Serving tens of thousands of domains over HTTPS with Caddy](https://caddy.community/t/serving-tens-of-thousands-of-domains-over-https-with-caddy/11179)
-
-### Cluster operations
-
-These endpoints have a dedicated OpenAPI spec.
- - APIv1 - [HTML spec](https://garagehq.deuxfleurs.fr/api/garage-admin-v1.html) - [OpenAPI YAML](https://garagehq.deuxfleurs.fr/api/garage-admin-v1.yml)
- - APIv0 (deprecated) - [HTML spec](https://garagehq.deuxfleurs.fr/api/garage-admin-v0.html) - [OpenAPI YAML](https://garagehq.deuxfleurs.fr/api/garage-admin-v0.yml)
-
-Requesting the API from the command line can be as simple as running:
-
-```bash
-curl -H 'Authorization: Bearer s3cr3t' http://localhost:3903/v0/status | jq
-```
-
-For more advanced use cases, we recommend using a SDK.  
-[Go to the "Build your own app" section to know how to use our SDKs](@/documentation/build/_index.md)
@@ -24,6 +24,7 @@ db_engine = "lmdb"

 block_size = "1M"
 block_ram_buffer_max = "256MiB"
+block_max_concurrent_reads = 16

 lmdb_map_size = "1T"

@@ -46,6 +47,7 @@ bootstrap_peers = [
    "212fd62eeaca72c122b45a7f4fa0f55e012aa5e24ac384a72a3016413fa724ff@[fc00:F::1]:3901",
 ]

+allow_punycode = false

 [consul_discovery]
 api = "catalog"
@@ -93,29 +95,31 @@ The following gives details about each available configuration option.

 [Environment variables](#env_variables).

-Top-level configuration options:
+Top-level configuration options, in alphabetical order:
+[`allow_punycode`](#allow_punycode),
 [`allow_world_readable_secrets`](#allow_world_readable_secrets),
+[`block_max_concurrent_reads`](`block_max_concurrent_reads),
 [`block_ram_buffer_max`](#block_ram_buffer_max),
 [`block_size`](#block_size),
 [`bootstrap_peers`](#bootstrap_peers),
 [`compression_level`](#compression_level),
+[`consistency_mode`](#consistency_mode),
 [`data_dir`](#data_dir),
 [`data_fsync`](#data_fsync),
 [`db_engine`](#db_engine),
 [`disable_scrub`](#disable_scrub),
-[`use_local_tz`](#use_local_tz),
 [`lmdb_map_size`](#lmdb_map_size),
 [`metadata_auto_snapshot_interval`](#metadata_auto_snapshot_interval),
 [`metadata_dir`](#metadata_dir),
 [`metadata_fsync`](#metadata_fsync),
 [`metadata_snapshots_dir`](#metadata_snapshots_dir),
 [`replication_factor`](#replication_factor),
-[`consistency_mode`](#consistency_mode),
 [`rpc_bind_addr`](#rpc_bind_addr),
 [`rpc_bind_outgoing`](#rpc_bind_outgoing),
 [`rpc_public_addr`](#rpc_public_addr),
 [`rpc_public_addr_subnet`](#rpc_public_addr_subnet)
-[`rpc_secret`/`rpc_secret_file`](#rpc_secret).
+[`rpc_secret`/`rpc_secret_file`](#rpc_secret),
+[`use_local_tz`](#use_local_tz).

 The `[consul_discovery]` section:
 [`api`](#consul_api),
@@ -153,13 +157,17 @@ The `[admin]` section:

 ### Environment variables {#env_variables}

-The following configuration parameter must be specified as an environment
-variable, it does not exist in the configuration file:
+The following configuration parameters must be specified as environment variables,
+they do not exist in the configuration file:

 - `GARAGE_LOG_TO_SYSLOG` (since `v0.9.4`): set this to `1` or `true` to make the
  Garage daemon send its logs to `syslog` (using the libc `syslog` function)
  instead of printing to stderr.

+- `GARAGE_LOG_TO_JOURNALD` (since `v1.2.0`): set this to `1` or `true` to make the
+  Garage daemon send its logs to `journald` (using the native protocol of `systemd-journald`)
+  instead of printing to stderr.
+
 The following environment variables can be used to override the corresponding
 values in the configuration file:

@@ -171,7 +179,7 @@ values in the configuration file:

 ### Top-level configuration options

-#### `replication_factor` {#replication_factor}
+#### `replication_factor` (since `v1.0.0`) {#replication_factor}

 The replication factor can be any positive integer smaller or equal the node count in your cluster.
 The chosen replication factor has a big impact on the cluster's failure tolerancy and performance characteristics.
@@ -219,7 +227,7 @@ is in progress.  In theory, no data should be lost as rebalancing is a
 routine operation for Garage, although we cannot guarantee you that everything
 will go right in such an extreme scenario.

-#### `consistency_mode` {#consistency_mode}
+#### `consistency_mode` (since `v1.0.0`) {#consistency_mode}

 The consistency mode setting determines the read and write behaviour of your cluster.

@@ -329,6 +337,7 @@ Since `v0.8.0`, Garage can use alternative storage backends as follows:
 | --------- | ----------------- | ------------- |
 | [LMDB](https://www.symas.com/lmdb) (since `v0.8.0`, default since `v0.9.0`) | `"lmdb"` | `<metadata_dir>/db.lmdb/` |
 | [Sqlite](https://sqlite.org) (since `v0.8.0`) | `"sqlite"` | `<metadata_dir>/db.sqlite` |
+| [Fjall](https://github.com/fjall-rs/fjall) (**experimental support** since `v1.3.0`) | `"fjall"` | `<metadata_dir>/db.fjall/` |
 | [Sled](https://sled.rs) (old default, removed since `v1.0`) | `"sled"` | `<metadata_dir>/db/` |

 Sled was supported until Garage v0.9.x, and was removed in Garage v1.0.
@@ -365,6 +374,14 @@ LMDB works very well, but is known to have the following limitations:
  so it is not the best choice for high-performance storage clusters,
  but it should work fine in many cases.

+- Fjall: a storage engine based on LSM trees, which theoretically allow for
+  higher write throughput than other storage engines that are based on B-trees.
+  Using Fjall could potentially improve Garage's performance significantly in
+  write-heavy workloads. **Support for Fjall is experimental at this point**,
+  we have added it to Garage for evaluation purposes only. **Do not use it for
+  production-critical workloads.**
+
+
 It is possible to convert Garage's metadata directory from one format to another
 using the `garage convert-db` command, which should be used as follows:

@@ -402,6 +419,7 @@ Here is how this option impacts the different database engines:
 |----------|------------------------------------|-------------------------------|
 | Sqlite   | `PRAGMA synchronous = OFF`         | `PRAGMA synchronous = NORMAL` |
 | LMDB     | `MDB_NOMETASYNC` + `MDB_NOSYNC`    | `MDB_NOMETASYNC`              |
+| Fjall    | default options                    | not supported                 |

 Note that the Sqlite database is always ran in `WAL` mode (`PRAGMA journal_mode = WAL`).

@@ -508,6 +526,29 @@ node.

 The default value is 256MiB.

+#### `block_max_concurrent_reads` (since `v1.3.0` / `v2.1.0`) {#block_max_concurrent_reads}
+
+The maximum number of blocks (individual files in the data directory) open
+simultaneously for reading.
+
+Reducing this number does not limit the number of data blocks that can be
+transferred through the network simultaneously. This mechanism was just added
+as a backpressure mechanism for HDD read speed: it helps avoid a situation
+where too many requests are coming in and Garage is reading too many block
+files simultaneously, thus not making timely progress on any of the reads.
+
+When a request to read a data block comes in through the network, the requests
+awaits for one of the `block_max_concurrent_reads` slots to be available
+(internally implemented using a Semaphore object). Once it acquired a read
+slot, it reads the entire block file to RAM and frees the slot as soon as the
+block file is finished reading. Only after the slot is released will the
+block's data start being transferred over the network.  If the request fails to
+acquire a reading slot wihtin 15 seconds, it fails with a timeout error.
+Timeout events can be monitored through the `block_read_semaphore_timeouts`
+metric in Prometheus: a non-zero number of such events indicates an I/O
+bottleneck on HDD read speed.
+
+
 #### `lmdb_map_size` {#lmdb_map_size}

 This parameters can be used to set the map size used by LMDB,
@@ -606,7 +647,7 @@ be obtained by running `garage node id` and then included directly in the
 key will be returned by `garage node id` and you will have to add the IP
 yourself.

-### `allow_world_readable_secrets` or `GARAGE_ALLOW_WORLD_READABLE_SECRETS` (env) {#allow_world_readable_secrets}
+#### `allow_world_readable_secrets` or `GARAGE_ALLOW_WORLD_READABLE_SECRETS` (env) {#allow_world_readable_secrets}

 Garage checks the permissions of your secret files to make sure they're not
 world-readable. In some cases, the check might fail and consider your files as
@@ -618,6 +659,13 @@ permission verification.
 Alternatively, you can set the `GARAGE_ALLOW_WORLD_READABLE_SECRETS`
 environment variable to `true` to bypass the permissions check.

+#### `allow_punycode` {#allow_punycode}
+
+Allow creating buckets with names containing punycode. When used for buckets served
+as websites, this allows using almost any unicode character in the domain name.
+
+Default to `false`.
+
 ### The `[consul_discovery]` section

 Garage supports discovering other nodes of the cluster using Consul.  For this
@@ -129,5 +129,5 @@ related to objects stored in an S3 bucket.
 In the context of our research project, [Aérogramme](https://aerogramme.deuxfleurs.fr),
 K2V is used to provide metadata and log storage for operations on encrypted e-mail storage.

-Learn more on the specification of K2V [here](https://git.deuxfleurs.fr/Deuxfleurs/garage/src/branch/k2v/doc/drafts/k2v-spec.md)
+Learn more on the specification of K2V [here](https://git.deuxfleurs.fr/Deuxfleurs/garage/src/commit/f8be15c37db857e177d543de7be863692628d567/doc/drafts/k2v-spec.md)
 and on how to enable it in Garage [here](@/documentation/reference-manual/k2v.md).
@@ -16,7 +16,7 @@ the `k2v` feature flag enabled can be obtained from our download page under
 with `-k2v` (example: `v0.7.2-k2v`).

 The specification of the K2V API can be found
-[here](https://git.deuxfleurs.fr/Deuxfleurs/garage/src/branch/main/doc/drafts/k2v-spec.md).
+[here](https://git.deuxfleurs.fr/Deuxfleurs/garage/src/commit/f8be15c37db857e177d543de7be863692628d567/doc/drafts/k2v-spec.md).
 This document also includes a high-level overview of K2V's design.

 The K2V API uses AWSv4 signatures for authentification, same as the S3 API.
@@ -23,7 +23,6 @@ Feel free to open a PR to suggest fixes this table. Minio is missing because the
 - 2022-05-25 - Many Ceph S3 endpoints are not documented but implemented. Following a notification from the Ceph community, we added them.


-
 ## High-level features

 | Feature                      | Garage                           | [Openstack Swift](https://docs.openstack.org/swift/latest/s3_compat.html) | [Ceph Object Gateway](https://docs.ceph.com/en/latest/radosgw/s3/) | [Riak CS](https://docs.riak.com/riak/cs/2.1.1/references/apis/storage/s3/index.html) | [OpenIO](https://docs.openio.io/latest/source/arch-design/s3_compliancy.html) |
@@ -34,6 +33,7 @@ Feel free to open a PR to suggest fixes this table. Minio is missing because the
 | [URL vhost-style](https://docs.aws.amazon.com/AmazonS3/latest/userguide/VirtualHosting.html#virtual-hosted-style-access) URL (eg. `bucket.host.tld/key`) |  ✅ Implemented | ❌| ✅| ✅ | ✅ |
 | [Presigned URLs](https://docs.aws.amazon.com/AmazonS3/latest/userguide/ShareObjectPreSignedURL.html) |  ✅ Implemented | ❌|  ✅ | ✅ |  ✅(❓) |
 | [SSE-C encryption](https://docs.aws.amazon.com/AmazonS3/latest/userguide/ServerSideEncryptionCustomerKeys.html) |  ✅ Implemented | ❓ |  ✅ | ❌ |  ✅ |
+| [Bucket versioning](https://docs.aws.amazon.com/AmazonS3/latest/userguide/Versioning.html) | ❌ Missing | ✅ |  ✅ | ❌ | ✅ |

 *Note:* OpenIO does not says if it supports presigned URLs. Because it is part
 of signature v4 and they claim they support it without additional precisions,
@@ -1,6 +1,6 @@
 +++
 title = "Migrating from 0.3 to 0.4"
-weight = 20
+weight = 80
 +++

 **Migrating from 0.3 to 0.4 is unsupported. This document is only intended to
@@ -1,6 +1,6 @@
 +++
 title = "Migrating from 0.5 to 0.6"
-weight = 15
+weight = 75
 +++

 **This guide explains how to migrate to 0.6 if you have an existing 0.5 cluster.
@@ -1,6 +1,6 @@
 +++
 title = "Migrating from 0.6 to 0.7"
-weight = 14
+weight = 74
 +++
 **This guide explains how to migrate to 0.7 if you have an existing 0.6 cluster.
 We don't recommend trying to migrate to 0.7 directly from 0.5 or older.**
@@ -1,6 +1,6 @@
 +++
 title = "Migrating from 0.7 to 0.8"
-weight = 13
+weight = 73
 +++

 **This guide explains how to migrate to 0.8 if you have an existing 0.7 cluster.
@@ -1,6 +1,6 @@
 +++
 title = "Migrating from 0.8 to 0.9"
-weight = 12
+weight = 72
 +++

 **This guide explains how to migrate to 0.9 if you have an existing 0.8 cluster.
@@ -1,6 +1,6 @@
 +++
 title = "Migrating from 0.9 to 1.0"
-weight = 11
+weight = 71
 +++

 **This guide explains how to migrate to 1.0 if you have an existing 0.9 cluster.
@@ -0,0 +1,70 @@
+++
+title = "Migrating from 1.0 to 2.0"
+weight = 70
+++
+
+**This guide explains how to migrate to v2.x if you have an existing v1.x.x cluster.
+We don't recommend trying to migrate to v2.x directly from v0.9.x or older.**
+
+This migration procedure has been tested on several clusters without issues.
+However, it is still a *critical procedure* that might cause issues.
+**Make sure to back up all your data before attempting it!**
+
+You might also want to read our [general documentation on upgrading Garage](@/documentation/operations/upgrading.md).
+
+## Changes introduced in v2.0
+
+The following are **breaking changes** in Garage v2.0 that require your attention when migrating:
+
+- The administration API has been completely reworked.
+  Some calls to the `/v1/` endpoints will still work but most will not.
+  New endpoints are prefixed by `/v2/`. **You will need to update all your code that makes use of the admin API.**
+
+- `replication_mode` is no longer a supported configuration parameter,
+  please use `replication_factor` and `consistency_mode` instead.
+
+## Migration procedure
+
+The migration to Garage v2.0 can be done with almost no downtime,
+by restarting all nodes at once in the new version.
+
+The migration steps are as follows:
+
+1. Do a `garage repair --all-nodes --yes tables`, check the logs and check that
+   all data seems to be synced correctly between nodes. If you have time, do
+   additional `garage repair` procedures (`blocks`, `versions`, `block_refs`,
+   etc.)
+
+2. Ensure you have a snapshot of your Garage installation that you can restore
+   to in case the upgrade goes wrong, with one of the following options:
+
+   - You may use the `garage meta snapshot --all` command
+     to make a backup snapshot of the metadata directories of your nodes
+     for backup purposes. Once this command has completed, copy the following
+     files and directories from the `metadata_dir` of all your nodes
+     to somewhere safe: `snapshots`, `cluster_layout`, `data_layout`,
+     `node_key`, `node_key.pub`. (If you have set the `metadata_snapshots_dir`
+     to a different value in your config file, back up that directory instead.)
+
+   - If you are running a filesystem such as ZFS or BTRFS that support
+     snapshotting, you can create a filesystem-level snapshot of the `metadata_dir`
+     of all your nodes to be used as a restoration point if needed.
+
+   - You may also make a back-up manually: turn off each node
+     individually; back up its metadata folder (for instance, use the following
+     command if your metadata directory is `/var/lib/garage/meta`: `cd
+     /var/lib/garage ; tar -acf meta-v1.0.tar.zst meta/`); turn it back on
+     again.  This will allow you to take a backup of all nodes without
+     impacting global cluster availability.  You can do all nodes of a single
+     zone at once as this does not impact the availability of Garage.
+
+3. Prepare your updated binaries and configuration files for Garage v2.0.
+   **Remember to update your configuration file to remove `replication_mode` and replace it by `replication_factor`.**
+
+4. Shut down all v1.0 nodes simultaneously, and restart them all simultaneously
+   in v2.0.  Use your favorite deployment tool (Ansible, Kubernetes, Nomad) to
+   achieve this as fast as possible.  Garage v2.0 should be in a working state
+   as soon as enough nodes have started.
+
+5. Monitor your cluster in the following hours to see if it works well under
+   your production load.
@@ -1,6 +1,6 @@
 +++
 title = "Testing strategy"
-weight = 30
+weight = 100
 +++


@@ -15,7 +15,10 @@ when changes are introduced.

 The Garage administration API was introduced in version 0.7.2, and was
 changed several times.
-This document applies only to the Garage v2 API (starting with Garage v2.0.0).
+
+**THIS DOCUMENT IS DEPRECATED.** We now have an OpenAPI spec which is automatically generated
+from Garage's source code and is always up-to-date. See `doc/api/garage-admin-v2.html`.
+Text in this document is no longer kept in sync with the admin API's actual behavior.


 ## Access control
@@ -53,6 +53,9 @@
          tests-sqlite = testWith {
            GARAGE_TEST_INTEGRATION_DB_ENGINE = "sqlite";
          };
+          tests-fjall = testWith {
+            GARAGE_TEST_INTEGRATION_DB_ENGINE = "fjall";
+          };
        };

        # ---- developpment shell, for making native builds only ----
@@ -78,6 +81,9 @@
                cargo-outdated
                cargo-machete
                nixpkgs-fmt
+                openssl
+                socat
+                killall
              ];
            };
          };
@@ -68,12 +68,13 @@ let
  rootFeatures = if features != null then
    features
  else
-    ([ "bundled-libs" "lmdb" "sqlite" "k2v" ] ++ (lib.optionals release [
+    ([ "bundled-libs" "lmdb" "sqlite" "fjall" "k2v" ] ++ (lib.optionals release [
      "consul-discovery"
      "kubernetes-discovery"
      "metrics"
      "telemetry-otlp"
      "syslog"
+      "journald"
    ]));

  featuresStr = lib.concatStringsSep "," rootFeatures;
@@ -30,6 +30,12 @@ for count in $(seq 1 3); do
 CONF_PATH="/tmp/config.$count.toml"
 LABEL="\e[${FANCYCOLORS[$count]}[$count]\e[49m"

+if [ "$GARAGE_OLDVER" == "v08" ]; then
+	REPLICATION_MODE="replication_mode = \"3\""
+else
+	REPLICATION_MODE="replication_factor = 3"
+fi
+
 cat > $CONF_PATH <<EOF
 block_size = 1048576			# objects are split in blocks of maximum this number of bytes
 metadata_dir = "/tmp/garage-meta-$count"
@@ -38,7 +44,7 @@ data_dir = "/tmp/garage-data-$count"
 rpc_bind_addr = "0.0.0.0:$((3900+$count))"		# the port other Garage nodes will use to talk to this node
 rpc_public_addr = "127.0.0.1:$((3900+$count))"
 bootstrap_peers = []
-replication_mode = "3"
+$REPLICATION_MODE
 rpc_secret = "$NETWORK_SECRET"

 [s3_api]
@@ -1,24 +1,18 @@
 apiVersion: v2
 name: garage
 description: S3-compatible object store for small self-hosted geo-distributed deployments
-
-# A chart can be either an 'application' or a 'library' chart.
-#
-# Application charts are a collection of templates that can be packaged into versioned archives
-# to be deployed.
-#
-# Library charts provide useful utilities or functions for the chart developer. They're included as
-# a dependency of application charts to inject those utilities and functions into the rendering
-# pipeline. Library charts do not define any templates and therefore cannot be deployed.
 type: application
+version: 0.9.1
+appVersion: "v2.1.0"
+home: https://garagehq.deuxfleurs.fr/
+icon: https://garagehq.deuxfleurs.fr/images/garage-logo.svg

-# This is the chart version. This version number should be incremented each time you make changes
-# to the chart and its templates, including the app version.
-# Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.8.0
+keywords:
+- geo-distributed
+- read-after-write-consistency
+- s3-compatible

-# This is the version number of the application being deployed. This version number should be
-# incremented each time you make changes to the application. Versions are not expected to
-# follow Semantic Versioning. They should reflect the version the application is using.
-# It is recommended to use it with quotes.
-appVersion: "v2.0.0"
+sources:
+- https://git.deuxfleurs.fr/Deuxfleurs/garage.git
+
+maintainers: []
@@ -1,14 +1,21 @@
 # garage

-![Version: 0.6.0](https://img.shields.io/badge/Version-0.6.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: v1.0.1](https://img.shields.io/badge/AppVersion-v1.0.1-informational?style=flat-square)
+![Version: 0.9.1](https://img.shields.io/badge/Version-0.9.1-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: v2.1.0](https://img.shields.io/badge/AppVersion-v2.1.0-informational?style=flat-square)

 S3-compatible object store for small self-hosted geo-distributed deployments

+**Homepage:** <https://garagehq.deuxfleurs.fr/>
+
+## Source Code
+
+* <https://git.deuxfleurs.fr/Deuxfleurs/garage.git>
+
 ## Values

 | Key | Type | Default | Description |
 |-----|------|---------|-------------|
 | affinity | object | `{}` |  |
+| commonLabels  | object | `{}` | Extra labels for all resources |
 | deployment.kind | string | `"StatefulSet"` | Switchable to DaemonSet |
 | deployment.podManagementPolicy | string | `"OrderedReady"` | If using statefulset, allow Parallel or OrderedReady (default) |
 | deployment.replicaCount | int | `3` | Number of StatefulSet replicas/garage nodes to start |
@@ -16,14 +23,16 @@ S3-compatible object store for small self-hosted geo-distributed deployments
 | extraVolumeMounts | object | `{}` |  |
 | extraVolumes | object | `{}` |  |
 | fullnameOverride | string | `""` |  |
-| garage.blockSize | string | `"1048576"` | Defaults is 1MB An increase can result in better performance in certain scenarios https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#block-size |
+| garage.blockSize | string | `"1048576"` | Defaults is 1MB An increase can result in better performance in certain scenarios https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#block_size |
 | garage.bootstrapPeers | list | `[]` | This is not required if you use the integrated kubernetes discovery |
-| garage.compressionLevel | string | `"1"` | zstd compression level of stored blocks https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#compression-level |
-| garage.dbEngine | string | `"lmdb"` | Can be changed for better performance on certain systems https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#db-engine-since-v0-8-0 |
+| garage.compressionLevel | string | `"1"` | zstd compression level of stored blocks https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#compression_level |
+| garage.dbEngine | string | `"lmdb"` | Can be changed for better performance on certain systems https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#db_engine |
 | garage.existingConfigMap | string | `""` | if not empty string, allow using an existing ConfigMap for the garage.toml, if set, ignores garage.toml |
 | garage.garageTomlString | string | `""` | String Template for the garage configuration if set, ignores above values. Values can be templated, see https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/ |
 | garage.kubernetesSkipCrd | bool | `false` | Set to true if you want to use k8s discovery but install the CRDs manually outside of the helm chart, for example if you operate at namespace level without cluster ressources |
-| garage.replicationMode | string | `"3"` | Default to 3 replicas, see the replication_mode section at https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#replication-mode |
+| garage.replicationFactor | string | `"3"` | Default to 3 replicas, see the replication_factor section at https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#replication_factor |
+| garage.consistencyMode | string | `"consistent"` | Default to read-after-write consistency, see the consistency_mode section at https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#consistency_mode |
+| garage.metadataAutoSnapshotInterval | string | `""` | If this value is set, Garage will automatically take a snapshot of the metadata DB file at a regular interval and save it in the metadata directory. https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#metadata_auto_snapshot_interval |
 | garage.rpcBindAddr | string | `"[::]:3901"` |  |
 | garage.rpcSecret | string | `""` | If not given, a random secret will be generated and stored in a Secret object |
 | garage.s3.api.region | string | `"garage"` |  |
@@ -49,6 +58,7 @@ S3-compatible object store for small self-hosted geo-distributed deployments
 | initImage.pullPolicy | string | `"IfNotPresent"` |  |
 | initImage.repository | string | `"busybox"` |  |
 | initImage.tag | string | `"stable"` |  |
+| livenessProbe | object | `{}` | Specifies a livenessProbe |
 | monitoring.metrics.enabled | bool | `false` | If true, a service for monitoring is created with a prometheus.io/scrape annotation |
 | monitoring.metrics.serviceMonitor.enabled | bool | `false` | If true, a ServiceMonitor CRD is created for a prometheus operator https://github.com/coreos/prometheus-operator |
 | monitoring.metrics.serviceMonitor.interval | string | `"15s"` |  |
@@ -71,6 +81,7 @@ S3-compatible object store for small self-hosted geo-distributed deployments
 | podSecurityContext.runAsGroup | int | `1000` |  |
 | podSecurityContext.runAsNonRoot | bool | `true` |  |
 | podSecurityContext.runAsUser | int | `1000` |  |
+| readinessProbe | object | `{}` | Specifies a readinessProbe |
 | resources | object | `{}` |  |
 | securityContext.capabilities | object | `{"drop":["ALL"]}` | The default security context is heavily restricted, feel free to tune it to your requirements |
 | securityContext.readOnlyRootFilesystem | bool | `true` |  |
@@ -47,6 +47,9 @@ helm.sh/chart: {{ include "garage.chart" . }}
 app.kubernetes.io/version: {{ .Chart.AppVersion | quote }}
 {{- end }}
 app.kubernetes.io/managed-by: {{ .Release.Service }}
+{{ with .Values.commonLabels }}
+{{- toYaml . }}
+{{- end }}
 {{- end }}

 {{/*
@@ -15,10 +15,15 @@ data:

    block_size = {{ .Values.garage.blockSize }}

-    replication_mode = "{{ .Values.garage.replicationMode }}"
+    replication_factor = {{ .Values.garage.replicationFactor }}
+    consistency_mode = "{{ .Values.garage.consistencyMode }}"

    compression_level = {{ .Values.garage.compressionLevel }}

+    {{- if .Values.garage.metadataAutoSnapshotInterval }}
+    metadata_auto_snapshot_interval = {{ .Values.garage.metadataAutoSnapshotInterval | quote }}
+    {{- end }}
+
    rpc_bind_addr = "{{ .Values.garage.rpcBindAddr }}"
    # rpc_secret will be populated by the init container from a k8s secret object
    rpc_secret = "__RPC_SECRET_REPLACE__"
@@ -21,7 +21,7 @@ spec:
        {{- toYaml . | nindent 8 }}
      {{- end }}
      labels:
-        {{- include "garage.selectorLabels" . | nindent 8 }}
+        {{- include "garage.labels" . | nindent 8 }}
    spec:
      {{- with .Values.imagePullSecrets }}
      imagePullSecrets:
@@ -78,15 +78,14 @@ spec:
          {{- with .Values.extraVolumeMounts }}
            {{- toYaml . | nindent 12 }}
          {{- end }}
-          # TODO
-          # livenessProbe:
-          #   httpGet:
-          #     path: /
-          #     port: 3900
-          # readinessProbe:
-          #   httpGet:
-          #     path: /
-          #     port: 3900
+          {{- with .Values.livenessProbe }}
+          livenessProbe:
+            {{- toYaml . | nindent 12 }}
+          {{- end }}
+          {{- with .Values.readinessProbe }}
+          readinessProbe:
+            {{- toYaml . | nindent 12 }}
+          {{- end }}
          resources:
            {{- toYaml .Values.resources | nindent 12 }}
      volumes:
@@ -2,25 +2,38 @@
 # This is a YAML-formatted file.
 # Declare variables to be passed into your templates.

+# -- Additional labels to add to all resources created by this chart
+commonLabels: {}
+#   app.kubernetes.io/part-of: storage
+#   team: platform
+
 # Garage configuration. These values go to garage.toml
 garage:
  # -- Can be changed for better performance on certain systems
-  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#db-engine-since-v0-8-0
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#db_engine
  dbEngine: "lmdb"

  # -- Defaults is 1MB
  # An increase can result in better performance in certain scenarios
-  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#block-size
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#block_size
  blockSize: "1048576"

-  # -- Default to 3 replicas, see the replication_mode section at
-  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#replication-mode
-  replicationMode: "3"
+  # -- Default to 3 replicas, see the replication_factor section at
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#replication_factor
+  replicationFactor: "3"
+
+  # -- By default, enable read-after-write consistency guarantees, see the consistency_mode section at
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#consistency_mode
+  consistencyMode: "consistent"

  # -- zstd compression level of stored blocks
-  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#compression-level
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#compression_level
  compressionLevel: "1"

+  # -- If this value is set, Garage will automatically take a snapshot of the metadata DB file at a regular interval and save it in the metadata directory.
+  # https://garagehq.deuxfleurs.fr/documentation/reference-manual/configuration/#metadata_auto_snapshot_interval
+  metadataAutoSnapshotInterval: ""
+
  rpcBindAddr: "[::]:3901"
  # -- If not given, a random secret will be generated and stored in a Secret object
  rpcSecret: ""
@@ -191,6 +204,21 @@ resources: {}
  #   cpu: 100m
  #   memory: 512Mi

+# -- Specifies a livenessProbe
+livenessProbe: {}
+  #httpGet:
+  #  path: /health
+  #  port: 3903
+  #initialDelaySeconds: 5
+  #periodSeconds: 30
+# -- Specifies a readinessProbe
+readinessProbe: {}
+  #httpGet:
+  #  path: /health
+  #  port: 3903
+  #initialDelaySeconds: 5
+  #periodSeconds: 30
+
 nodeSelector: {}

 tolerations: []
@@ -43,7 +43,7 @@
             "rpc_bind_addr = \"0.0.0.0:3901\"\n"
             "rpc_public_addr = \"" node ":3901\"\n"
             "db_engine = \"lmdb\"\n"
-             "replication_mode = \"3\"\n"
+             "replication_factor = 3\n"
             "data_dir = \"" data-dir "\"\n"
             "metadata_dir = \"" meta-dir "\"\n"
             "[s3_api]\n"
@@ -8,7 +8,7 @@ data:
    metadata_dir = "/tmp/meta"
    data_dir = "/tmp/data"

-    replication_mode = "3"
+    replication_factor = 3

    rpc_bind_addr = "[::]:3901"
    rpc_secret = "1799bccfd7411eddcf9ebd316bc1f5287ad12a68094e1c6ac6abde7e6feae1ec"
@@ -0,0 +1,43 @@
+apiVersion: apiextensions.k8s.io/v1
+kind: CustomResourceDefinition
+metadata:
+  name: garagenodes.deuxfleurs.fr
+spec:
+  conversion:
+    strategy: None
+  group: deuxfleurs.fr
+  names:
+    kind: GarageNode
+    listKind: GarageNodeList
+    plural: garagenodes
+    singular: garagenode
+  scope: Namespaced
+  versions:
+  - name: v1
+    schema:
+      openAPIV3Schema:
+        description: Auto-generated derived type for Node via `CustomResource`
+        properties:
+          spec:
+            properties:
+              address:
+                format: ip
+                type: string
+              hostname:
+                type: string
+              port:
+                format: uint16
+                minimum: 0
+                type: integer
+            required:
+            - address
+            - hostname
+            - port
+            type: object
+        required:
+        - spec
+        title: GarageNode
+        type: object
+    served: true
+    storage: true
+    subresources: {}
@@ -0,0 +1,5 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+resources:
+- garagenodes.deuxfleurs.fr.yaml
@@ -694,32 +694,7 @@
            ]
          }
        },
-        "overrides": [
-          {
-            "__systemRef": "hideSeriesFrom",
-            "matcher": {
-              "id": "byNames",
-              "options": {
-                "mode": "exclude",
-                "names": [
-                  "10.83.2.3:3903"
-                ],
-                "prefix": "All except:",
-                "readOnly": true
-              }
-            },
-            "properties": [
-              {
-                "id": "custom.hideFrom",
-                "value": {
-                  "legend": false,
-                  "tooltip": false,
-                  "viz": true
-                }
-              }
-            ]
-          }
-        ]
+        "overrides": []
      },
      "gridPos": {
        "h": 8,
@@ -119,8 +119,12 @@ if [ -z "$SKIP_BOTO3" ]; then
  AWS_ENDPOINT_URL=https://localhost:4443 python <<EOF
 import boto3
 client = boto3.client('s3', verify=False)
+print("Put&delete hello world object")
 client.put_object(Body=b'hello world', Bucket='eprouvette', Key='test.s3.txt')
 client.delete_object(Bucket='eprouvette', Key='test.s3.txt')
+print("Put&delete big object")
+client.upload_file("/tmp/garage.3.rnd", 'eprouvette', 'garage.3.rnd')
+client.delete_object(Bucket='eprouvette', Key='garage.3.rnd')
 print("OK!")
 EOF
 fi
@@ -37,6 +37,7 @@ in
    ];
    shellHook = ''
      function to_s3 {
+        AWS_REQUEST_CHECKSUM_CALCULATION=WHEN_REQUIRED AWS_RESPONSE_CHECKSUM_VALIDATION=WHEN_REQUIRED \
        aws \
            --endpoint-url https://garage.deuxfleurs.fr \
            --region garage \
@@ -93,6 +94,7 @@ in

        nix-build nix/build_index.nix

+        AWS_REQUEST_CHECKSUM_CALCULATION=WHEN_REQUIRED AWS_RESPONSE_CHECKSUM_VALIDATION=WHEN_REQUIRED \
        aws \
            --endpoint-url https://garage.deuxfleurs.fr \
            --region garage \
@@ -100,6 +102,7 @@ in
            result/share/_releases.json \
            s3://garagehq.deuxfleurs.fr/

+        AWS_REQUEST_CHECKSUM_CALCULATION=WHEN_REQUIRED AWS_RESPONSE_CHECKSUM_VALIDATION=WHEN_REQUIRED \
        aws \
            --endpoint-url https://garage.deuxfleurs.fr \
            --region garage \
@@ -1,6 +1,6 @@
 [package]
 name = "garage_api_admin"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -175,6 +175,63 @@ impl RequestHandler for DeleteAdminTokenRequest {
 	}
 }

+impl RequestHandler for GetCurrentAdminTokenInfoRequest {
+	type Response = GetCurrentAdminTokenInfoResponse;
+
+	async fn handle(
+		self,
+		garage: &Arc<Garage>,
+		_admin: &Admin,
+	) -> Result<GetCurrentAdminTokenInfoResponse, Error> {
+		let now = now_msec();
+
+		if garage
+			.config
+			.admin
+			.metrics_token
+			.as_ref()
+			.is_some_and(|s| s == &self.admin_token)
+		{
+			return Ok(GetCurrentAdminTokenInfoResponse(
+				GetAdminTokenInfoResponse {
+					id: None,
+					created: None,
+					name: "metrics_token (from daemon configuration)".into(),
+					expiration: None,
+					expired: false,
+					scope: vec!["Metrics".into()],
+				},
+			));
+		}
+
+		if garage
+			.config
+			.admin
+			.admin_token
+			.as_ref()
+			.is_some_and(|s| s == &self.admin_token)
+		{
+			return Ok(GetCurrentAdminTokenInfoResponse(
+				GetAdminTokenInfoResponse {
+					id: None,
+					created: None,
+					name: "admin_token (from daemon configuration)".into(),
+					expiration: None,
+					expired: false,
+					scope: vec!["*".into()],
+				},
+			));
+		}
+
+		let (prefix, _) = self.admin_token.split_once('.').unwrap();
+		let token = get_existing_admin_token(&garage, &prefix.to_string()).await?;
+
+		Ok(GetCurrentAdminTokenInfoResponse(admin_token_info_results(
+			&token, now,
+		)))
+	}
+}
+
 // ---- helpers ----

 fn admin_token_info_results(token: &AdminApiToken, now: u64) -> GetAdminTokenInfoResponse {
@@ -12,7 +12,7 @@ use garage_rpc::*;

 use garage_model::garage::Garage;

-use garage_api_common::helpers::is_default;
+use garage_api_common::{common_error::CommonError, helpers::is_default};

 use crate::api_server::{find_matching_nodes, AdminRpc, AdminRpcResponse};
 use crate::error::Error;
@@ -56,6 +56,7 @@ admin_endpoints![
 	CreateAdminToken,
 	UpdateAdminToken,
 	DeleteAdminToken,
+	GetCurrentAdminTokenInfo,

 	// Layout operations
 	GetClusterLayout,
@@ -144,6 +145,13 @@ pub struct MultiResponse<RB> {
 	pub error: HashMap<String, String>,
 }

+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
+pub struct MultiRequestQueryParams {
+	/// Node ID to query, or `*` for all nodes, or `self` for the node responding to the request
+	pub node: String,
+}
+
 // **********************************************
 //      Special endpoints
 //
@@ -154,8 +162,10 @@ pub struct MultiResponse<RB> {
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct OptionsRequest;

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct CheckDomainRequest {
+	/// The domain name to check for
 	pub domain: String,
 }

@@ -257,7 +267,7 @@ pub struct GetClusterHealthResponse {
 	/// the number of storage nodes currently registered in the cluster layout
 	pub storage_nodes: usize,
 	/// the number of storage nodes to which a connection is currently open
-	pub storage_nodes_ok: usize,
+	pub storage_nodes_up: usize,
 	/// the total number of partitions of the data (currently always 256)
 	pub partitions: usize,
 	/// the number of partitions for which a quorum of write nodes is available
@@ -308,6 +318,7 @@ pub struct ListAdminTokensResponse(pub Vec<GetAdminTokenInfoResponse>);
 // ---- GetAdminTokenInfo ----

 #[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 #[serde(rename_all = "camelCase")]
 pub struct GetAdminTokenInfoRequest {
 	/// Admin API token ID
@@ -353,9 +364,12 @@ pub struct CreateAdminTokenResponse {

 // ---- UpdateAdminToken ----

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct UpdateAdminTokenRequest {
+	/// Admin API token ID
 	pub id: String,
+	#[param(ignore = true)]
 	pub body: UpdateAdminTokenRequestBody,
 }

@@ -382,14 +396,25 @@ pub struct UpdateAdminTokenResponse(pub GetAdminTokenInfoResponse);

 // ---- DeleteAdminToken ----

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct DeleteAdminTokenRequest {
+	/// Admin API token ID
 	pub id: String,
 }

 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct DeleteAdminTokenResponse;

+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct GetCurrentAdminTokenInfoRequest {
+	pub admin_token: String,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize, ToSchema)]
+#[serde(rename_all = "camelCase")]
+pub struct GetCurrentAdminTokenInfoResponse(pub GetAdminTokenInfoResponse);
+
 // **********************************************
 //      Layout operations
 // **********************************************
@@ -648,6 +673,7 @@ pub struct ListKeysResponseItem {
 // ---- GetKeyInfo ----

 #[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 #[serde(rename_all = "camelCase")]
 pub struct GetKeyInfoRequest {
 	/// Access key ID
@@ -724,9 +750,12 @@ pub struct ImportKeyResponse(pub GetKeyInfoResponse);

 // ---- UpdateKey ----

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct UpdateKeyRequest {
+	/// Access key ID
 	pub id: String,
+	#[param(ignore = true)]
 	pub body: UpdateKeyRequestBody,
 }

@@ -751,8 +780,10 @@ pub struct UpdateKeyRequestBody {

 // ---- DeleteKey ----

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct DeleteKeyRequest {
+	/// Access key ID
 	pub id: String,
 }

@@ -790,6 +821,7 @@ pub struct BucketLocalAlias {
 // ---- GetBucketInfo ----

 #[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 #[serde(rename_all = "camelCase")]
 pub struct GetBucketInfoRequest {
 	/// Exact bucket ID to look up
@@ -878,9 +910,12 @@ pub struct CreateBucketLocalAlias {

 // ---- UpdateBucket ----

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct UpdateBucketRequest {
+	/// ID of the bucket to update
 	pub id: String,
+	#[param(ignore = true)]
 	pub body: UpdateBucketRequestBody,
 }

@@ -904,8 +939,10 @@ pub struct UpdateBucketWebsiteAccess {

 // ---- DeleteBucket ----

-#[derive(Debug, Clone, Serialize, Deserialize)]
+#[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 pub struct DeleteBucketRequest {
+	/// ID of the bucket to delete
 	pub id: String,
 }

@@ -928,6 +965,7 @@ pub struct CleanupIncompleteUploadsResponse {
 }

 #[derive(Debug, Clone, Serialize, Deserialize, IntoParams)]
+#[into_params(parameter_in = Query)]
 #[serde(rename_all = "camelCase")]
 pub struct InspectObjectRequest {
 	pub bucket_id: String,
@@ -1114,6 +1152,8 @@ pub enum RepairType {
 	BlockRc,
 	Rebalance,
 	Scrub(ScrubCommand),
+	Aliases,
+	ClearResyncQueue,
 }

 #[derive(Debug, Clone, Serialize, Deserialize, ToSchema)]
@@ -1308,4 +1348,5 @@ pub struct LocalPurgeBlocksResponse {
 	pub objects_deleted: u64,
 	pub uploads_deleted: u64,
 	pub versions_deleted: u64,
+	pub block_refs_purged: u64,
 }
@@ -217,6 +217,13 @@ impl ApiHandler for ArcAdminApiServer {
 	) -> Result<Response<ResBody>, Error> {
 		self.0.handle_http_api(req, endpoint).await
 	}
+
+	fn key_id_from_request(&self, req: &Request<IncomingBody>) -> Option<String> {
+		let auth_header = req.headers().get(AUTHORIZATION)?;
+		let token = parse_authorization(auth_header).ok()?;
+		let key_id = token.split_once('.')?.0;
+		Some(key_id.to_string())
+	}
 }

 impl ApiEndpoint for HttpEndpoint {
@@ -244,6 +251,15 @@ fn hash_bearer_token(token: &str) -> String {
 		.to_string()
 }

+fn parse_authorization(auth_header: &hyper::http::HeaderValue) -> Result<&str, Error> {
+	let token = auth_header
+		.to_str()?
+		.strip_prefix("Bearer ")
+		.ok_or_else(|| Error::forbidden("Invalid Authorization header"))?
+		.trim();
+	Ok(token)
+}
+
 fn verify_authorization(
 	garage: &Garage,
 	global_token_hash: Option<&str>,
@@ -260,11 +276,7 @@ fn verify_authorization(
 				"Bearer token must be provided in Authorization header",
 			))
 		}
-		Some(authorization) => authorization
-			.to_str()?
-			.strip_prefix("Bearer ")
-			.ok_or_else(|| Error::forbidden("Invalid Authorization header"))?
-			.trim(),
+		Some(authorization) => parse_authorization(authorization)?,
 	};

 	let token_hash_string = if let Some((prefix, _)) = token.split_once('.') {
@@ -273,7 +285,8 @@ fn verify_authorization(
 			.get_local(&EmptyKey, &prefix.to_string())?
 			.and_then(|k| k.state.into_option())
 			.filter(|p| !p.is_expired(now_msec()))
-			.filter(|p| p.has_scope(endpoint_name))
+			// GetCurrentAdminTokenInfo endpoint must be accessible even if it is not in the token scopes
+			.filter(|p| p.has_scope(endpoint_name) || endpoint_name == "GetCurrentAdminTokenInfo")
 			.ok_or_else(|| Error::forbidden(invalid_msg))?
 			.token_hash
 	} else {
@@ -293,38 +306,36 @@ fn verify_authorization(
 }

 pub(crate) fn find_matching_nodes(garage: &Garage, spec: &str) -> Result<Vec<Uuid>, Error> {
-	let mut res = vec![];
-	if spec == "*" {
-		res = garage.system.cluster_layout().all_nodes().to_vec();
+	if spec == "self" {
+		Ok(vec![garage.system.id])
+	} else {
+		// Collect all nodes currently up and/or in cluster layout
+		let mut res = vec![];
+		if let Ok(all_nodes) = garage.system.cluster_layout().all_nodes() {
+			res = all_nodes.to_vec();
+		}
 		for node in garage.system.get_known_nodes() {
 			if node.is_up && !res.contains(&node.id) {
 				res.push(node.id);
 			}
 		}
-	} else if spec == "self" {
-		res.push(garage.system.id);
-	} else {
-		let layout = garage.system.cluster_layout();
-		let known_nodes = garage.system.get_known_nodes();
-		let all_nodes = layout
-			.all_nodes()
-			.iter()
-			.copied()
-			.chain(known_nodes.iter().filter(|x| x.is_up).map(|x| x.id));
-		for node in all_nodes {
-			if !res.contains(&node) && hex::encode(node).starts_with(spec) {
-				res.push(node);
+
+		if spec == "*" {
+			// match all nodes
+			Ok(res)
+		} else {
+			// filter nodes that match spec
+			res.retain(|node| hex::encode(node).starts_with(spec));
+			if res.is_empty() {
+				Err(Error::bad_request(format!("No nodes matching {}", spec)))
+			} else if res.len() > 1 {
+				Err(Error::bad_request(format!(
+					"Multiple nodes matching {}: {:?}",
+					spec, res
+				)))
+			} else {
+				Ok(res)
 			}
 		}
-		if res.is_empty() {
-			return Err(Error::bad_request(format!("No nodes matching {}", spec)));
-		}
-		if res.len() > 1 {
-			return Err(Error::bad_request(format!(
-				"Multiple nodes matching {}: {:?}",
-				spec, res
-			)));
-		}
 	}
-	Ok(res)
 }
@@ -151,6 +151,7 @@ impl RequestHandler for LocalPurgeBlocksRequest {
 		let mut obj_dels = 0;
 		let mut mpu_dels = 0;
 		let mut ver_dels = 0;
+		let mut br_dels = 0;

 		for hash in self.0.iter() {
 			let hash = hex::decode(hash).ok_or_bad_request("invalid hash")?;
@@ -176,11 +177,18 @@ impl RequestHandler for LocalPurgeBlocksRequest {
 						ver_dels += 1;
 					}
 				}
+				if !br.deleted.get() {
+					let mut br = br;
+					br.deleted.set();
+					garage.block_ref_table.insert(&br).await?;
+					br_dels += 1;
+				}
 			}
 		}

 		Ok(LocalPurgeBlocksResponse {
 			blocks_purged: self.0.len() as u64,
+			block_refs_purged: br_dels,
 			versions_deleted: ver_dels,
 			objects_deleted: obj_dels,
 			uploads_deleted: mpu_dels,
@@ -159,7 +159,7 @@ impl RequestHandler for CreateBucketRequest {
 		let helper = garage.locked_helper().await;

 		if let Some(ga) = &self.global_alias {
-			if !is_valid_bucket_name(ga) {
+			if !is_valid_bucket_name(ga, garage.config.allow_punycode) {
 				return Err(Error::bad_request(format!(
 					"{}: {}",
 					ga, INVALID_BUCKET_NAME_MESSAGE
@@ -174,7 +174,7 @@ impl RequestHandler for CreateBucketRequest {
 		}

 		if let Some(la) = &self.local_alias {
-			if !is_valid_bucket_name(&la.alias) {
+			if !is_valid_bucket_name(&la.alias, garage.config.allow_punycode) {
 				return Err(Error::bad_request(format!(
 					"{}: {}",
 					la.alias, INVALID_BUCKET_NAME_MESSAGE
@@ -255,7 +255,7 @@ impl RequestHandler for DeleteBucketRequest {
 		for ((key_id, alias), _, active) in state.local_aliases.items().iter() {
 			if *active {
 				helper
-					.unset_local_bucket_alias(bucket.id, key_id, alias)
+					.purge_local_bucket_alias(bucket.id, key_id, alias)
 					.await?;
 			}
 		}
@@ -697,20 +697,21 @@ async fn bucket_info_results(
 		}),
 		keys: relevant_keys
 			.into_values()
-			.map(|key| {
+			.filter_map(|key| {
 				let p = key.state.as_option().unwrap();
-				GetBucketInfoKey {
+				let permissions = p
+					.authorized_buckets
+					.get(&bucket.id)
+					.filter(|p| p.is_any())
+					.map(|p| ApiBucketKeyPerm {
+						read: p.allow_read,
+						write: p.allow_write,
+						owner: p.allow_owner,
+					})?;
+				Some(GetBucketInfoKey {
 					access_key_id: key.key_id,
 					name: p.name.get().to_string(),
-					permissions: p
-						.authorized_buckets
-						.get(&bucket.id)
-						.map(|p| ApiBucketKeyPerm {
-							read: p.allow_read,
-							write: p.allow_write,
-							owner: p.allow_owner,
-						})
-						.unwrap_or_default(),
+					permissions,
 					bucket_local_aliases: p
 						.local_aliases
 						.items()
@@ -718,7 +719,7 @@ async fn bucket_info_results(
 						.filter(|(_, _, b)| *b == Some(bucket.id))
 						.map(|(n, _, _)| n.to_string())
 						.collect::<Vec<_>>(),
-				}
+				})
 			})
 			.collect::<Vec<_>>(),
 		objects: *counters.get(OBJECTS).unwrap_or(&0),
@@ -56,48 +56,52 @@ impl RequestHandler for GetClusterStatusRequest {
 			})
 			.collect::<HashMap<_, _>>();

-		for (id, _, role) in layout.current().roles.items().iter() {
-			if let layout::NodeRoleV(Some(r)) = role {
-				let role = NodeAssignedRole {
-					zone: r.zone.to_string(),
-					capacity: r.capacity,
-					tags: r.tags.clone(),
-				};
-				match nodes.get_mut(id) {
-					None => {
-						nodes.insert(
-							*id,
-							NodeResp {
-								id: hex::encode(id),
-								role: Some(role),
-								..Default::default()
-							},
-						);
-					}
-					Some(n) => {
-						n.role = Some(role);
+		if let Ok(current_layout) = layout.current() {
+			for (id, _, role) in current_layout.roles.items().iter() {
+				if let layout::NodeRoleV(Some(r)) = role {
+					let role = NodeAssignedRole {
+						zone: r.zone.to_string(),
+						capacity: r.capacity,
+						tags: r.tags.clone(),
+					};
+					match nodes.get_mut(id) {
+						None => {
+							nodes.insert(
+								*id,
+								NodeResp {
+									id: hex::encode(id),
+									role: Some(role),
+									..Default::default()
+								},
+							);
+						}
+						Some(n) => {
+							n.role = Some(role);
+						}
 					}
 				}
 			}
 		}

-		for ver in layout.versions().iter().rev().skip(1) {
-			for (id, _, role) in ver.roles.items().iter() {
-				if let layout::NodeRoleV(Some(r)) = role {
-					if r.capacity.is_some() {
-						if let Some(n) = nodes.get_mut(id) {
-							if n.role.is_none() {
-								n.draining = true;
+		if let Ok(layout_versions) = layout.versions() {
+			for ver in layout_versions.iter().rev().skip(1) {
+				for (id, _, role) in ver.roles.items().iter() {
+					if let layout::NodeRoleV(Some(r)) = role {
+						if r.capacity.is_some() {
+							if let Some(n) = nodes.get_mut(id) {
+								if n.role.is_none() {
+									n.draining = true;
+								}
+							} else {
+								nodes.insert(
+									*id,
+									NodeResp {
+										id: hex::encode(id),
+										draining: true,
+										..Default::default()
+									},
+								);
 							}
-						} else {
-							nodes.insert(
-								*id,
-								NodeResp {
-									id: hex::encode(id),
-									draining: true,
-									..Default::default()
-								},
-							);
 						}
 					}
 				}
@@ -108,7 +112,7 @@ impl RequestHandler for GetClusterStatusRequest {
 		nodes.sort_by(|x, y| x.id.cmp(&y.id));

 		Ok(GetClusterStatusResponse {
-			layout_version: layout.current().version,
+			layout_version: layout.inner().current().version,
 			nodes,
 		})
 	}
@@ -134,7 +138,9 @@ impl RequestHandler for GetClusterHealthRequest {
 			known_nodes: health.known_nodes,
 			connected_nodes: health.connected_nodes,
 			storage_nodes: health.storage_nodes,
-			storage_nodes_ok: health.storage_nodes_ok,
+			// Translating storage_nodes_up (admin API context) to storage_nodes_ok (metrics context)
+			// TODO: when releasing major release, consider renaming all the fields in the metrics to storage_nodes_up
+			storage_nodes_up: health.storage_nodes_ok,
 			partitions: health.partitions,
 			partitions_quorum: health.partitions_quorum,
 			partitions_all_ok: health.partitions_all_ok,
@@ -157,9 +163,11 @@ impl RequestHandler for GetClusterStatisticsRequest {
 		// Gather storage node and free space statistics for current nodes
 		let layout = &garage.system.cluster_layout();
 		let mut node_partition_count = HashMap::<Uuid, u64>::new();
-		for short_id in layout.current().ring_assignment_data.iter() {
-			let id = layout.current().node_id_vec[*short_id as usize];
-			*node_partition_count.entry(id).or_default() += 1;
+		if let Ok(current_layout) = layout.current() {
+			for short_id in current_layout.ring_assignment_data.iter() {
+				let id = current_layout.node_id_vec[*short_id as usize];
+				*node_partition_count.entry(id).or_default() += 1;
+			}
 		}
 		let node_info = garage
 			.system
@@ -172,7 +180,11 @@ impl RequestHandler for GetClusterStatisticsRequest {
 		for (id, parts) in node_partition_count.iter() {
 			let info = node_info.get(id);
 			let status = info.map(|x| &x.status);
-			let role = layout.current().roles.get(id).and_then(|x| x.0.as_ref());
+			let role = layout
+				.current()
+				.ok()
+				.and_then(|l| l.roles.get(id))
+				.and_then(|x| x.0.as_ref());
 			let hostname = status.and_then(|x| x.hostname.as_deref()).unwrap_or("?");
 			let zone = role.map(|x| x.zone.as_str()).unwrap_or("?");
 			let capacity = role
@@ -95,6 +95,7 @@ impl ApiError for Error {
 	fn add_http_headers(&self, header_map: &mut HeaderMap<HeaderValue>) {
 		use hyper::header;
 		header_map.append(header::CONTENT_TYPE, "application/json".parse().unwrap());
+		header_map.append(header::ACCESS_CONTROL_ALLOW_ORIGIN, "*".parse().unwrap());
 	}

 	fn http_body(&self, garage_region: &str, path: &str) -> ErrorBody {
@@ -230,9 +230,18 @@ async fn key_info_results(
 		},
 		buckets: relevant_buckets
 			.into_values()
-			.map(|bucket| {
+			.filter_map(|bucket| {
 				let state = bucket.state.as_option().unwrap();
-				KeyInfoBucketResponse {
+				let permissions = key_state
+					.authorized_buckets
+					.get(&bucket.id)
+					.filter(|p| p.is_any())
+					.map(|p| ApiBucketKeyPerm {
+						read: p.allow_read,
+						write: p.allow_write,
+						owner: p.allow_owner,
+					})?;
+				Some(KeyInfoBucketResponse {
 					id: hex::encode(bucket.id),
 					global_aliases: state
 						.aliases
@@ -248,16 +257,8 @@ async fn key_info_results(
 						.filter(|((k, _), _, a)| *a && *k == key.key_id)
 						.map(|((_, n), _, _)| n.to_string())
 						.collect::<Vec<_>>(),
-					permissions: key_state
-						.authorized_buckets
-						.get(&bucket.id)
-						.map(|p| ApiBucketKeyPerm {
-							read: p.allow_read,
-							write: p.allow_write,
-							owner: p.allow_owner,
-						})
-						.unwrap_or_default(),
-				}
+					permissions,
+				})
 			})
 			.collect::<Vec<_>>(),
 	};
@@ -74,16 +74,18 @@ macro_rules! admin_endpoints {
                type Response = AdminApiResponse;

                async fn handle(self, garage: &Arc<Garage>, admin: &Admin) -> Result<AdminApiResponse, Error> {
-                    Ok(match self {
+                    match self {
                        $(
-                            AdminApiRequest::$special_endpoint(_) => panic!(
-                                concat!(stringify!($special_endpoint), " needs to go through a special handler")
+                            AdminApiRequest::$special_endpoint(_) => Err(
+                                Error::Common(CommonError::BadRequest(
+                                    concat!(stringify!($special_endpoint), " cannot be used outside of the HTTP Admin API").into()
+                                ))
                            ),
                        )*
                        $(
-                            AdminApiRequest::$endpoint(req) => AdminApiResponse::$endpoint(req.handle(garage, admin).await?),
+                            AdminApiRequest::$endpoint(req) => Ok(AdminApiResponse::$endpoint(req.handle(garage, admin).await?)),
                        )*
-                    })
+                    }
                }
            }
        }
@@ -106,17 +106,17 @@ impl RequestHandler for LocalGetNodeStatisticsRequest {

 		// Gather block manager statistics
 		writeln!(&mut ret, "\nBlock manager stats:").unwrap();
-		let rc_len = garage.block_manager.rc_len()?.to_string();
+		let rc_len = garage.block_manager.rc_approximate_len()?.to_string();

 		ret += &format_table_to_string(vec![
 			format!("  number of RC entries:\t{} (~= number of blocks)", rc_len),
 			format!(
 				"  resync queue length:\t{}",
-				garage.block_manager.resync.queue_len()?
+				garage.block_manager.resync.queue_approximate_len()?
 			),
 			format!(
 				"  blocks with resync errors:\t{}",
-				garage.block_manager.resync.errors_len()?
+				garage.block_manager.resync.errors_approximate_len()?
 			),
 		]);

@@ -129,16 +129,21 @@ where
 	F: TableSchema + 'static,
 	R: TableReplication + 'static,
 {
-	let data_len = t.data.store.len().map_err(GarageError::from)?.to_string();
-	let mkl_len = t.merkle_updater.merkle_tree_len()?.to_string();
+	let data_len = t
+		.data
+		.store
+		.approximate_len()
+		.map_err(GarageError::from)?
+		.to_string();
+	let mkl_len = t.merkle_updater.merkle_tree_approximate_len()?.to_string();

 	Ok(format!(
 		"  {}\t{}\t{}\t{}\t{}\t{}",
 		F::TABLE_NAME,
 		data_len,
 		mkl_len,
-		t.merkle_updater.todo_len()?,
-		t.data.insert_queue_len()?,
-		t.data.gc_todo_len()?
+		t.merkle_updater.todo_approximate_len()?,
+		t.data.insert_queue_approximate_len()?,
+		t.data.gc_todo_approximate_len()?
 	))
 }
@@ -46,9 +46,7 @@ a static website for the requested domain. This is used by reverse proxies such
 as Caddy or Tricot, to avoid requesting TLS certificates for domain names that
 do not correspond to an actual website.
    ",
-    params(
-        ("domain", description = "The domain name to check for"),
-    ),
+    params(CheckDomainRequest),
    security(()),
 	responses(
            (status = 200, description = "The domain name redirects to a static website bucket"),
@@ -167,9 +165,7 @@ fn CreateAdminToken() -> () {}
 Updates information about the specified admin API token.
    ",
    request_body = UpdateAdminTokenRequestBody,
-    params(
-        ("id", description = "Admin API token ID"),
-    ),
+    params(UpdateAdminTokenRequest),
 	responses(
            (status = 200, description = "Admin token has been updated", body = UpdateAdminTokenResponse),
            (status = 500, description = "Internal server error")
@@ -181,9 +177,7 @@ fn UpdateAdminToken() -> () {}
    path = "/v2/DeleteAdminToken",
    tag = "Admin API token",
    description = "Delete an admin API token from the cluster, revoking all its permissions.",
-    params(
-        ("id", description = "Admin API token ID"),
-    ),
+    params(DeleteAdminTokenRequest),
 	responses(
            (status = 200, description = "Admin token has been deleted"),
            (status = 500, description = "Internal server error")
@@ -191,6 +185,19 @@ fn UpdateAdminToken() -> () {}
 )]
 fn DeleteAdminToken() -> () {}

+#[utoipa::path(get,
+    path = "/v2/GetCurrentAdminTokenInfo",
+    tag = "Admin API token",
+    description = "
+Return information about the calling admin API token.
+    ",
+    responses(
+            (status = 200, description = "Information about the admin token", body = GetCurrentAdminTokenInfoResponse),
+            (status = 500, description = "Internal server error")
+    ),
+)]
+fn GetCurrentAdminTokenInfo() -> () {}
+
 // **********************************************
 //      Layout operations
 // **********************************************
@@ -378,9 +385,7 @@ Updates information about the specified API access key.
 *Note: the secret key is not returned in the response, `null` is sent instead.*
    ",
    request_body = UpdateKeyRequestBody,
-    params(
-        ("id", description = "Access key ID"),
-    ),
+    params(UpdateKeyRequest),
 	responses(
            (status = 200, description = "Access key has been updated", body = UpdateKeyResponse),
            (status = 500, description = "Internal server error")
@@ -392,9 +397,7 @@ fn UpdateKey() -> () {}
    path = "/v2/DeleteKey",
    tag = "Access key",
    description = "Delete a key from the cluster. Its access will be removed from all the buckets. Buckets are not automatically deleted and can be dangling. You should manually delete them before. ",
-    params(
-        ("id", description = "Access key ID"),
-    ),
+    params(DeleteKeyRequest),
 	responses(
            (status = 200, description = "Access key has been deleted"),
            (status = 500, description = "Internal server error")
@@ -465,9 +468,7 @@ In `quotas`: new values of `maxSize` and `maxObjects` must both be specified, or
 to remove the quotas. An absent value will be considered the same as a `null`. It is not possible
 to change only one of the two quotas.
    ",
-    params(
-        ("id", description = "ID of the bucket to update"),
-    ),
+    params(UpdateBucketRequest),
    request_body = UpdateBucketRequestBody,
 	responses(
            (status = 200, description = "Bucket has been updated", body = UpdateBucketResponse),
@@ -485,9 +486,7 @@ Deletes a storage bucket. A bucket cannot be deleted if it is not empty.

 **Warning:** this will delete all aliases associated with the bucket!
    ",
-    params(
-        ("id", description = "ID of the bucket to delete"),
-    ),
+    params(DeleteBucketRequest),
 	responses(
            (status = 200, description = "Bucket has been deleted"),
            (status = 400, description = "Bucket is not empty"),
@@ -617,9 +616,7 @@ fn RemoveBucketAlias() -> () {}
    description = "
 Return information about the Garage daemon running on one or several nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalGetNodeInfoResponse>),
            (status = 500, description = "Internal server error")
@@ -635,9 +632,7 @@ Fetch statistics for one or several Garage nodes.

 *Note: do not try to parse the `freeform` field of the response, it is given as a string specifically because its format is not stable.*
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalGetNodeStatisticsResponse>),
            (status = 500, description = "Internal server error")
@@ -651,9 +646,7 @@ fn GetNodeStatistics() -> () {}
    description = "
 Instruct one or several nodes to take a snapshot of their metadata databases.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalCreateMetadataSnapshotResponse>),
            (status = 500, description = "Internal server error")
@@ -667,9 +660,7 @@ fn CreateMetadataSnapshot() -> () {}
    description = "
 Launch a repair operation on one or several cluster nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalLaunchRepairOperationRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalLaunchRepairOperationResponse>),
@@ -688,9 +679,7 @@ fn LaunchRepairOperation() -> () {}
    description = "
 List background workers currently running on one or several cluster nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalListWorkersRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalListWorkersResponse>),
@@ -705,9 +694,7 @@ fn ListWorkers() -> () {}
    description = "
 Get information about the specified background worker on one or several cluster nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalGetWorkerInfoRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalGetWorkerInfoResponse>),
@@ -722,9 +709,7 @@ fn GetWorkerInfo() -> () {}
    description = "
 Fetch values of one or several worker variables, from one or several cluster nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalGetWorkerVariableRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalGetWorkerVariableResponse>),
@@ -739,9 +724,7 @@ fn GetWorkerVariable() -> () {}
    description = "
 Set the value for a worker variable, on one or several cluster nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalSetWorkerVariableRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalSetWorkerVariableResponse>),
@@ -760,9 +743,7 @@ fn SetWorkerVariable() -> () {}
    description = "
 List data blocks that are currently in an errored state on one or several Garage nodes.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalListBlockErrorsResponse>),
            (status = 500, description = "Internal server error")
@@ -776,9 +757,7 @@ fn ListBlockErrors() -> () {}
    description = "
 Get detailed information about a data block stored on a Garage node, including all object versions and in-progress multipart uploads that contain a reference to this block.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalGetBlockInfoRequest,
 	responses(
            (status = 200, description = "Detailed block information", body = MultiResponse<LocalGetBlockInfoResponse>),
@@ -793,9 +772,7 @@ fn GetBlockInfo() -> () {}
    description = "
 Instruct Garage node(s) to retry the resynchronization of one or several missing data block(s).
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalRetryBlockResyncRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalRetryBlockResyncResponse>),
@@ -812,9 +789,7 @@ Purge references to one or several missing data blocks.

 This will remove all objects and in-progress multipart uploads that contain the specified data block(s). The objects will be permanently deleted from the buckets in which they appear. Use with caution.
    ",
-    params(
-        ("node", description = "Node ID to query, or `*` for all nodes, or `self` for the node responding to the request"),
-    ),
+    params(MultiRequestQueryParams),
    request_body = LocalPurgeBlocksRequest,
 	responses(
            (status = 200, description = "Responses from individual cluster nodes", body = MultiResponse<LocalPurgeBlocksResponse>),
@@ -843,7 +818,7 @@ impl Modify for SecurityAddon {
 #[derive(OpenApi)]
 #[openapi(
    info(
-        version = "v2.0.0",
+        version = "v2.1.0",
        title = "Garage administration API",
        description = "Administrate your Garage cluster programatically, including status, layout, keys, buckets, and maintainance tasks.

@@ -872,6 +847,7 @@ impl Modify for SecurityAddon {
        CreateAdminToken,
        UpdateAdminToken,
        DeleteAdminToken,
+        GetCurrentAdminTokenInfo,
        // Layout operations
        GetClusterLayout,
        GetClusterLayoutHistory,
@@ -89,6 +89,19 @@ impl RequestHandler for LocalLaunchRepairOperationRequest {
 					garage.block_manager.clone(),
 				));
 			}
+			RepairType::Aliases => {
+				info!("Repairing bucket aliases (foreground)");
+				garage.locked_helper().await.repair_aliases().await?;
+			}
+			RepairType::ClearResyncQueue => {
+				info!("Clearing resync queue (foreground)");
+				let garage = garage.clone();
+				tokio::task::spawn_blocking(move || {
+					garage.block_manager.resync.clear_resync_queue()
+				})
+				.await
+				.map_err(garage_util::error::Error::from)??;
+			}
 		}
 		Ok(LocalLaunchRepairOperationResponse)
 	}
@@ -40,6 +40,7 @@ impl AdminApiRequest {
 			POST CreateAdminToken (body),
 			POST UpdateAdminToken (body_field, query::id),
 			POST DeleteAdminToken (query::id),
+			GET GetCurrentAdminTokenInfo (admin_token),
 			// Layout endpoints
 			GET GetClusterLayout (),
 			GET GetClusterLayoutHistory (),
@@ -1,6 +1,6 @@
 [package]
 name = "garage_api_common"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -21,15 +21,12 @@ garage_util.workspace = true
 base64.workspace = true
 bytes.workspace = true
 chrono.workspace = true
-crc32fast.workspace = true
-crc32c.workspace = true
-crc64fast-nvme.workspace = true
+crc-fast.workspace = true
 crypto-common.workspace = true
 err-derive.workspace = true
 hex.workspace = true
 hmac.workspace = true
 md-5.workspace = true
-idna.workspace = true
 tracing.workspace = true
 nom.workspace = true
 pin-project.workspace = true
@@ -59,6 +59,12 @@ pub trait ApiHandler: Send + Sync + 'static {
 		req: Request<IncomingBody>,
 		endpoint: Self::Endpoint,
 	) -> impl Future<Output = Result<Response<BoxBody<Self::Error>>, Self::Error>> + Send;
+
+	/// Returns the key id used to authenticate this request. The ID returned must be safe to
+	/// log.
+	fn key_id_from_request(&self, _req: &Request<IncomingBody>) -> Option<String> {
+		None
+	}
 }

 pub struct ApiServer<A: ApiHandler> {
@@ -143,19 +149,20 @@ impl<A: ApiHandler> ApiServer<A> {
 	) -> Result<Response<BoxBody<A::Error>>, http::Error> {
 		let uri = req.uri().clone();

-		if let Ok(forwarded_for_ip_addr) =
+		let source = if let Ok(forwarded_for_ip_addr) =
 			forwarded_headers::handle_forwarded_for_headers(req.headers())
 		{
-			info!(
-				"{} (via {}) {} {}",
-				forwarded_for_ip_addr,
-				addr,
-				req.method(),
-				uri
-			);
+			format!("{forwarded_for_ip_addr} (via {addr})")
 		} else {
-			info!("{} {} {}", addr, req.method(), uri);
-		}
+			format!("{addr}")
+		};
+		// we only do this to log the access key, so we can discard any error
+		let key = self
+			.api_handler
+			.key_id_from_request(&req)
+			.map(|k| format!("(key {k}) "))
+			.unwrap_or_default();
+		info!("{source} {key}{} {uri}", req.method());
 		debug!("{:?}", req);

 		let tracer = opentelemetry::global::tracer("garage");
@@ -344,7 +351,11 @@ where

 	while !*must_exit.borrow() {
 		let (stream, client_addr) = tokio::select! {
-			acc = listener.accept() => acc?,
+			acc = listener.accept() => match acc {
+							Ok(r) => r,
+							Err(e) if e.kind() == std::io::ErrorKind::ConnectionAborted => continue,
+							Err(e) => return Err(e.into()),
+						},
 			_ = must_exit.changed() => continue,
 		};

@@ -8,7 +8,6 @@ use hyper::{
 	body::{Body, Bytes},
 	Request, Response,
 };
-use idna::domain_to_unicode;
 use serde::{Deserialize, Serialize};

 use garage_model::bucket_table::BucketParams;
@@ -97,7 +96,7 @@ pub fn authority_to_host(authority: &str) -> Result<String, Error> {
 			authority
 		))),
 	};
-	authority.map(|h| domain_to_unicode(h).0)
+	authority.map(|h| h.to_ascii_lowercase())
 }

 /// Extract the bucket name and the key name from an HTTP path and possibly a bucket provided in
@@ -83,6 +83,21 @@ macro_rules! router_match {
            parse_json_body::< [<$api Request>], _, Error>($req).await?
        })
    }};
+   (@@gen_parse_request $api:ident, (admin_token), $query: expr, $req:expr) => {{
+        paste!({
+            let auth_header = $req.headers()
+                .get(hyper::header::AUTHORIZATION)
+                .ok_or_else(|| Error::bad_request("Missing Authorization header"))?
+                .to_str()
+                .map_err(|_| Error::bad_request("Invalid Authorization header"))?;
+
+            let admin_token = auth_header.strip_prefix("Bearer ")
+                .ok_or_else(|| Error::bad_request("Authorization header must be Bearer token"))?
+                .to_string();
+
+            [< $api Request >] { admin_token }
+        })
+    }};
    (@@gen_parse_request $api:ident, (body_field, $($conv:ident $(($conv_arg:expr))? :: $param:ident),*), $query: expr, $req:expr)
        =>
    {{
@@ -1,10 +1,7 @@
-use std::convert::{TryFrom, TryInto};
-use std::hash::Hasher;
+use std::convert::TryInto;

 use base64::prelude::*;
-use crc32c::Crc32cHasher as Crc32c;
-use crc32fast::Hasher as Crc32;
-use crc64fast_nvme::Digest as Crc64Nvme;
+use crc_fast::{CrcAlgorithm, Digest as CrcDigest};
 use md5::{Digest, Md5};
 use sha1::Sha1;
 use sha2::Sha256;
@@ -22,6 +19,7 @@ pub const CONTENT_MD5: HeaderName = HeaderName::from_static("content-md5");
 pub const X_AMZ_CHECKSUM_ALGORITHM: HeaderName =
 	HeaderName::from_static("x-amz-checksum-algorithm");
 pub const X_AMZ_CHECKSUM_MODE: HeaderName = HeaderName::from_static("x-amz-checksum-mode");
+pub const X_AMZ_CHECKSUM_TYPE: HeaderName = HeaderName::from_static("x-amz-checksum-type");
 pub const X_AMZ_CHECKSUM_CRC32: HeaderName = HeaderName::from_static("x-amz-checksum-crc32");
 pub const X_AMZ_CHECKSUM_CRC32C: HeaderName = HeaderName::from_static("x-amz-checksum-crc32c");
 pub const X_AMZ_CHECKSUM_CRC64NVME: HeaderName =
@@ -29,6 +27,10 @@ pub const X_AMZ_CHECKSUM_CRC64NVME: HeaderName =
 pub const X_AMZ_CHECKSUM_SHA1: HeaderName = HeaderName::from_static("x-amz-checksum-sha1");
 pub const X_AMZ_CHECKSUM_SHA256: HeaderName = HeaderName::from_static("x-amz-checksum-sha256");

+// Values for x-amz-checksum-type
+pub const COMPOSITE: &str = "COMPOSITE";
+pub const FULL_OBJECT: &str = "FULL_OBJECT";
+
 pub type Crc32Checksum = [u8; 4];
 pub type Crc32cChecksum = [u8; 4];
 pub type Crc64NvmeChecksum = [u8; 8];
@@ -36,6 +38,21 @@ pub type Md5Checksum = [u8; 16];
 pub type Sha1Checksum = [u8; 20];
 pub type Sha256Checksum = [u8; 32];

+// -- MAP OF CRC ALGORITHMS :
+// CRC32        -> CrcAlgorithm::Crc32IsoHdlc
+// CRC32C       -> CrcAlgorithm::Crc32Iscsi
+// CRC64NVME    -> CrcAlgorithm::Crc64Nvme
+
+pub fn new_crc32() -> CrcDigest {
+	CrcDigest::new(CrcAlgorithm::Crc32IsoHdlc)
+}
+pub fn new_crc32c() -> CrcDigest {
+	CrcDigest::new(CrcAlgorithm::Crc32Iscsi)
+}
+pub fn new_crc64nvme() -> CrcDigest {
+	CrcDigest::new(CrcAlgorithm::Crc64Nvme)
+}
+
 #[derive(Debug, Default, Clone)]
 pub struct ExpectedChecksums {
 	// base64-encoded md5 (content-md5 header)
@@ -47,9 +64,9 @@ pub struct ExpectedChecksums {
 }

 pub struct Checksummer {
-	pub crc32: Option<Crc32>,
-	pub crc32c: Option<Crc32c>,
-	pub crc64nvme: Option<Crc64Nvme>,
+	pub crc32: Option<CrcDigest>,
+	pub crc32c: Option<CrcDigest>,
+	pub crc64nvme: Option<CrcDigest>,
 	pub md5: Option<Md5>,
 	pub sha1: Option<Sha1>,
 	pub sha256: Option<Sha256>,
@@ -98,13 +115,13 @@ impl Checksummer {
 			self.sha256 = Some(Sha256::new());
 		}
 		if matches!(&expected.extra, Some(ChecksumValue::Crc32(_))) {
-			self.crc32 = Some(Crc32::new());
+			self.crc32 = Some(new_crc32());
 		}
 		if matches!(&expected.extra, Some(ChecksumValue::Crc32c(_))) {
-			self.crc32c = Some(Crc32c::default());
+			self.crc32c = Some(new_crc32c());
 		}
 		if matches!(&expected.extra, Some(ChecksumValue::Crc64Nvme(_))) {
-			self.crc64nvme = Some(Crc64Nvme::default());
+			self.crc64nvme = Some(new_crc64nvme());
 		}
 		if matches!(&expected.extra, Some(ChecksumValue::Sha1(_))) {
 			self.sha1 = Some(Sha1::new());
@@ -114,13 +131,13 @@ impl Checksummer {
 	pub fn add(mut self, algo: Option<ChecksumAlgorithm>) -> Self {
 		match algo {
 			Some(ChecksumAlgorithm::Crc32) => {
-				self.crc32 = Some(Crc32::new());
+				self.crc32 = Some(new_crc32());
 			}
 			Some(ChecksumAlgorithm::Crc32c) => {
-				self.crc32c = Some(Crc32c::default());
+				self.crc32c = Some(new_crc32c());
 			}
 			Some(ChecksumAlgorithm::Crc64Nvme) => {
-				self.crc64nvme = Some(Crc64Nvme::default());
+				self.crc64nvme = Some(new_crc64nvme());
 			}
 			Some(ChecksumAlgorithm::Sha1) => {
 				self.sha1 = Some(Sha1::new());
@@ -138,10 +155,10 @@ impl Checksummer {
 			crc32.update(bytes);
 		}
 		if let Some(crc32c) = &mut self.crc32c {
-			crc32c.write(bytes);
+			crc32c.update(bytes);
 		}
 		if let Some(crc64nvme) = &mut self.crc64nvme {
-			crc64nvme.write(bytes);
+			crc64nvme.update(bytes);
 		}
 		if let Some(md5) = &mut self.md5 {
 			md5.update(bytes);
@@ -156,11 +173,9 @@ impl Checksummer {

 	pub fn finalize(self) -> Checksums {
 		Checksums {
-			crc32: self.crc32.map(|x| u32::to_be_bytes(x.finalize())),
-			crc32c: self
-				.crc32c
-				.map(|x| u32::to_be_bytes(u32::try_from(x.finish()).unwrap())),
-			crc64nvme: self.crc64nvme.map(|x| u64::to_be_bytes(x.sum64())),
+			crc32: self.crc32.map(|x| u32::to_be_bytes(x.finalize() as u32)),
+			crc32c: self.crc32c.map(|x| u32::to_be_bytes(x.finalize() as u32)),
+			crc64nvme: self.crc64nvme.map(|x| u64::to_be_bytes(x.finalize())),
 			md5: self.md5.map(|x| x.finalize()[..].try_into().unwrap()),
 			sha1: self.sha1.map(|x| x.finalize()[..].try_into().unwrap()),
 			sha256: self.sha256.map(|x| x.finalize()[..].try_into().unwrap()),
@@ -192,10 +207,11 @@ impl Checksums {
 		}
 		if let Some(extra) = expected.extra {
 			let algo = extra.algorithm();
-			if self.extract(Some(algo)) != Some(extra) {
+			let calculated = self.extract(Some(algo));
+			if calculated != Some(extra) {
 				return Err(Error::InvalidDigest(format!(
-					"Failed to validate checksum for algorithm {:?}",
-					algo
+					"Failed to validate checksum for algorithm {:?}: calculated {:?}, expected {:?}",
+					algo, calculated, extra
 				)));
 			}
 		}
@@ -424,7 +424,7 @@ pub fn verify_v4(
 // ============ Authorization header, or X-Amz-* query params =========

 pub struct Authorization {
-	key_id: String,
+	pub key_id: String,
 	scope: String,
 	signed_headers: String,
 	signature: String,
@@ -433,7 +433,7 @@ pub struct Authorization {
 }

 impl Authorization {
-	fn parse_header(headers: &HeaderMap) -> Result<Self, Error> {
+	pub fn parse_header(headers: &HeaderMap) -> Result<Self, Error> {
 		let authorization = headers
 			.get(AUTHORIZATION)
 			.ok_or_bad_request("Missing authorization header")?
@@ -1,6 +1,6 @@
 [package]
 name = "garage_api_k2v"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -171,6 +171,12 @@ impl ApiHandler for K2VApiServer {

 		Ok(resp_ok)
 	}
+
+	fn key_id_from_request(&self, req: &Request<IncomingBody>) -> Option<String> {
+		garage_api_common::signature::payload::Authorization::parse_header(req.headers())
+			.map(|auth| auth.key_id)
+			.ok()
+	}
 }

 impl ApiEndpoint for K2VApiEndpoint {
@@ -99,6 +99,7 @@ impl ApiError for Error {
 	fn add_http_headers(&self, header_map: &mut HeaderMap<HeaderValue>) {
 		use hyper::header;
 		header_map.append(header::CONTENT_TYPE, "application/json".parse().unwrap());
+		header_map.append(header::ACCESS_CONTROL_ALLOW_ORIGIN, "*".parse().unwrap());
 	}

 	fn http_body(&self, garage_region: &str, path: &str) -> ErrorBody {
@@ -28,7 +28,7 @@ pub async fn handle_read_index(
 	let node_id_vec = garage
 		.system
 		.cluster_layout()
-		.all_nongateway_nodes()
+		.all_nongateway_nodes()?
 		.to_vec();

 	let (partition_keys, more, next_start) = read_range(
@@ -66,7 +66,7 @@ pub async fn handle_read_index(
 					bytes: *vals.get(&s_bytes).unwrap_or(&0),
 				}
 			})
-			.collect::<Vec<_>>(),
+			.collect(),
 		more,
 		next_start,
 	};
@@ -1,6 +1,6 @@
 [package]
 name = "garage_api_s3"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -27,9 +27,7 @@ async-compression.workspace = true
 base64.workspace = true
 bytes.workspace = true
 chrono.workspace = true
-crc32fast.workspace = true
-crc32c.workspace = true
-crc64fast-nvme.workspace = true
+crc-fast.workspace = true
 err-derive.workspace = true
 hex.workspace = true
 hmac.workspace = true
@@ -223,6 +223,7 @@ impl ApiHandler for S3ApiServer {
 			Endpoint::DeleteBucket {} => handle_delete_bucket(ctx).await,
 			Endpoint::GetBucketLocation {} => handle_get_bucket_location(ctx),
 			Endpoint::GetBucketVersioning {} => handle_get_bucket_versioning(),
+			Endpoint::GetBucketAcl {} => handle_get_bucket_acl(ctx),
 			Endpoint::ListObjects {
 				delimiter,
 				encoding_type,
@@ -339,6 +340,12 @@ impl ApiHandler for S3ApiServer {

 		Ok(resp_ok)
 	}
+
+	fn key_id_from_request(&self, req: &Request<IncomingBody>) -> Option<String> {
+		garage_api_common::signature::payload::Authorization::parse_header(req.headers())
+			.map(|auth| auth.key_id)
+			.ok()
+	}
 }

 impl ApiEndpoint for S3ApiEndpoint {
@@ -5,7 +5,7 @@ use hyper::{Request, Response, StatusCode};
 use garage_model::bucket_alias_table::*;
 use garage_model::bucket_table::Bucket;
 use garage_model::garage::Garage;
-use garage_model::key_table::Key;
+use garage_model::key_table::{Key, KeyParams};
 use garage_model::permission::BucketKeyPerm;
 use garage_table::util::*;
 use garage_util::crdt::*;
@@ -44,6 +44,55 @@ pub fn handle_get_bucket_versioning() -> Result<Response<ResBody>, Error> {
 		.body(string_body(xml))?)
 }

+pub fn handle_get_bucket_acl(ctx: ReqCtx) -> Result<Response<ResBody>, Error> {
+	let ReqCtx {
+		bucket_id, api_key, ..
+	} = ctx;
+	let key_p = api_key.params().ok_or_internal_error(
+		"Key should not be in deleted state at this point (in handle_get_bucket_acl)",
+	)?;
+
+	let mut grants: Vec<s3_xml::Grant> = vec![];
+	let kp = api_key.bucket_permissions(&bucket_id);
+
+	if kp.allow_owner {
+		grants.push(s3_xml::Grant {
+			grantee: create_grantee(&key_p, &api_key),
+			permission: s3_xml::Value("FULL_CONTROL".to_string()),
+		});
+	} else {
+		if kp.allow_read {
+			grants.push(s3_xml::Grant {
+				grantee: create_grantee(&key_p, &api_key),
+				permission: s3_xml::Value("READ".to_string()),
+			});
+			grants.push(s3_xml::Grant {
+				grantee: create_grantee(&key_p, &api_key),
+				permission: s3_xml::Value("READ_ACP".to_string()),
+			});
+		}
+		if kp.allow_write {
+			grants.push(s3_xml::Grant {
+				grantee: create_grantee(&key_p, &api_key),
+				permission: s3_xml::Value("WRITE".to_string()),
+			});
+		}
+	}
+
+	let access_control_policy = s3_xml::AccessControlPolicy {
+		xmlns: (),
+		owner: None,
+		acl: s3_xml::AccessControlList { entries: grants },
+	};
+
+	let xml = s3_xml::to_xml_with_header(&access_control_policy)?;
+	trace!("xml: {}", xml);
+
+	Ok(Response::builder()
+		.header("Content-Type", "application/xml")
+		.body(string_body(xml))?)
+}
+
 pub async fn handle_list_buckets(
 	garage: &Garage,
 	api_key: &Key,
@@ -167,7 +216,7 @@ pub async fn handle_create_bucket(
 		}

 		// Create the bucket!
-		if !is_valid_bucket_name(&bucket_name) {
+		if !is_valid_bucket_name(&bucket_name, garage.config.allow_punycode) {
 			return Err(Error::bad_request(format!(
 				"{}: {}",
 				bucket_name, INVALID_BUCKET_NAME_MESSAGE
@@ -236,11 +285,11 @@ pub async fn handle_delete_bucket(ctx: ReqCtx) -> Result<Response<ResBody>, Erro
 		// 1. delete bucket alias
 		if is_local_alias {
 			helper
-				.unset_local_bucket_alias(*bucket_id, &api_key.key_id, bucket_name)
+				.purge_local_bucket_alias(*bucket_id, &api_key.key_id, bucket_name)
 				.await?;
 		} else {
 			helper
-				.unset_global_bucket_alias(*bucket_id, bucket_name)
+				.purge_global_bucket_alias(*bucket_id, bucket_name)
 				.await?;
 		}

@@ -306,6 +355,15 @@ fn parse_create_bucket_xml(xml_bytes: &[u8]) -> Option<Option<String>> {
 	Some(ret)
 }

+fn create_grantee(key_params: &KeyParams, api_key: &Key) -> s3_xml::Grantee {
+	s3_xml::Grantee {
+		xmlns_xsi: (),
+		typ: "CanonicalUser".to_string(),
+		display_name: Some(s3_xml::Value(key_params.name.get().to_string())),
+		id: Some(s3_xml::Value(api_key.key_id.to_string())),
+	}
+}
+
 #[cfg(test)]
 mod tests {
 	use super::*;
@@ -26,9 +26,10 @@ use garage_api_common::signature::checksum::*;
 use crate::api_server::{ReqBody, ResBody};
 use crate::encryption::{EncryptionParams, OekDerivationInfo};
 use crate::error::*;
-use crate::get::{full_object_byte_stream, PreconditionHeaders};
+use crate::get::{check_version_not_deleted, full_object_byte_stream, PreconditionHeaders};
 use crate::multipart;
 use crate::put::{extract_metadata_headers, save_stream, ChecksumMode, SaveStreamResult};
+use crate::website::X_AMZ_WEBSITE_REDIRECT_LOCATION;
 use crate::xml::{self as s3_xml, xmlns_tag};

 pub const X_AMZ_COPY_SOURCE_IF_MATCH: HeaderName =
@@ -78,8 +79,24 @@ pub async fn handle_copy(
 		},
 	)?;

+	let was_multipart = source_version_meta.etag.contains('-') // HACK
+        || source_object_meta_inner.checksum_type == Some(ChecksumType::Composite);
+
 	// Extract source checksum info before source_object_meta_inner is consumed
 	let source_checksum = source_object_meta_inner.checksum;
+	let source_checksum_type = match (source_object_meta_inner.checksum_type, source_checksum) {
+		(Some(ct), _) => Some(ct),
+		(None, Some(_)) => {
+			// Migrated object from garage v1.x or older
+			// determine checksum type depending if this is a multipart upload or not
+			if was_multipart {
+				Some(ChecksumType::Composite)
+			} else {
+				Some(ChecksumType::FullObject)
+			}
+		}
+		(None, None) => None,
+	};
 	let source_checksum_algorithm = source_checksum.map(|x| x.algorithm());

 	// If source object has a checksum, the destination object must as well.
@@ -88,15 +105,26 @@ pub async fn handle_copy(
 	let checksum_algorithm = checksum_algorithm.or(source_checksum_algorithm);

 	// Determine metadata of destination object
-	let was_multipart = source_version_meta.etag.contains('-');
 	let dest_object_meta = ObjectVersionMetaInner {
 		headers: match req.headers().get("x-amz-metadata-directive") {
 			Some(v) if v == hyper::header::HeaderValue::from_static("REPLACE") => {
 				extract_metadata_headers(req.headers())?
 			}
-			_ => source_object_meta_inner.into_owned().headers,
+			_ => {
+				// The x-amz-website-redirect-location header is not copied, instead
+				// it is replaced by the value from the request (or removed if no
+				// value was specified)
+				let is_redirect =
+					|(key, _): &(String, String)| key == X_AMZ_WEBSITE_REDIRECT_LOCATION.as_str();
+				let mut headers: Vec<_> = source_object_meta_inner.headers.clone();
+				headers.retain(|h| !is_redirect(h));
+				let new_headers = extract_metadata_headers(req.headers())?;
+				headers.extend(new_headers.into_iter().filter(is_redirect));
+				headers
+			}
 		},
 		checksum: source_checksum,
+		checksum_type: source_checksum_type,
 	};

 	// Do actual object copying
@@ -116,8 +144,8 @@ pub async fn handle_copy(
 	// See: https://docs.aws.amazon.com/AmazonS3/latest/userguide/checking-object-integrity.html

 	let must_recopy = !EncryptionParams::is_same(&source_encryption, &dest_encryption)
-		|| source_checksum_algorithm != checksum_algorithm
-		|| (was_multipart && checksum_algorithm.is_some());
+		|| (checksum_algorithm.is_some()
+			&& (was_multipart || checksum_algorithm != source_checksum_algorithm));

 	let res = if !must_recopy {
 		// In most cases, we can just copy the metadata and link blocks of the
@@ -134,18 +162,21 @@ pub async fn handle_copy(
 		)
 		.await?
 	} else {
-		let expected_checksum = ExpectedChecksums {
-			md5: None,
-			sha256: None,
-			extra: source_checksum,
-		};
 		let checksum_mode = if was_multipart || source_checksum_algorithm != checksum_algorithm {
 			ChecksumMode::Calculate(checksum_algorithm)
 		} else {
-			ChecksumMode::Verify(&expected_checksum)
+			ChecksumMode::Verify(ExpectedChecksums {
+				md5: None,
+				sha256: None,
+				extra: source_checksum,
+			})
+		};
+		// For multipart uploads that had a composite checksum, set checksum type
+		// to full object as it will be recalculated.
+		let dest_object_meta = ObjectVersionMetaInner {
+			checksum_type: checksum_algorithm.map(|_| ChecksumType::FullObject),
+			..dest_object_meta
 		};
-		// If source and dest encryption use different keys,
-		// we must decrypt content and re-encrypt, so rewrite all data blocks.
 		handle_copy_reencrypt(
 			ctx,
 			dest_key,
@@ -237,6 +268,7 @@ async fn handle_copy_metaonly(
 				.get(&source_version.uuid, &EmptyKey)
 				.await?;
 			let source_version = source_version.ok_or(Error::NoSuchKey)?;
+			check_version_not_deleted(&source_version)?;

 			// Write an "uploading" marker in Object table
 			// This holds a reference to the object in the Version table
@@ -325,7 +357,7 @@ async fn handle_copy_reencrypt(
 	source_version: &ObjectVersion,
 	source_version_data: &ObjectVersionData,
 	source_encryption: EncryptionParams,
-	checksum_mode: ChecksumMode<'_>,
+	checksum_mode: ChecksumMode,
 ) -> Result<SaveStreamResult, Error> {
 	// basically we will read the source data (decrypt if necessary)
 	// and save that in a new object (encrypt if necessary),
@@ -437,6 +469,7 @@ pub async fn handle_upload_part_copy(
 		.get(&source_object_version.uuid, &EmptyKey)
 		.await?
 		.ok_or(Error::NoSuchKey)?;
+	check_version_not_deleted(&source_version)?;

 	// We want to reuse blocks from the source version as much as possible.
 	// However, we still need to get the data from these blocks
@@ -512,7 +545,7 @@ pub async fn handle_upload_part_copy(

 	// Now, actually copy the blocks
 	let mut checksummer = Checksummer::init(&Default::default(), !dest_encryption.is_encrypted())
-		.add(dest_object_checksum_algorithm);
+		.add(dest_object_checksum_algorithm.map(|(algo, _)| algo));

 	// First, create a stream that is able to read the source blocks
 	// and extract the subrange if necessary.
@@ -568,6 +601,7 @@ pub async fn handle_upload_part_copy(

 	let mut current_offset = 0;
 	let mut next_block = defragmenter.next().await?;
+	let mut blocks_to_dup = dest_version.clone();

 	// TODO this could be optimized similarly to read_and_put_blocks
 	// low priority because uploadpartcopy is rarely used
@@ -597,8 +631,7 @@ pub async fn handle_upload_part_copy(
 			.unwrap()?;
 		checksummer = checksummer_updated;

-		dest_version.blocks.clear();
-		dest_version.blocks.put(
+		let (version_block_key, version_block) = (
 			VersionBlockKey {
 				part_number,
 				offset: current_offset,
@@ -610,40 +643,59 @@ pub async fn handle_upload_part_copy(
 		);
 		current_offset += data_len;

-		let block_ref = BlockRef {
-			block: final_hash,
-			version: dest_version_id,
-			deleted: false.into(),
+		let next = if let Some(final_data) = data_to_upload {
+			dest_version.blocks.clear();
+			dest_version.blocks.put(version_block_key, version_block);
+			let block_ref = BlockRef {
+				block: final_hash,
+				version: dest_version_id,
+				deleted: false.into(),
+			};
+			let (_, _, _, next) = futures::try_join!(
+				// Thing 1: if the block is not exactly a block that existed before,
+				// we need to insert that data as a new block.
+				garage.block_manager.rpc_put_block(
+					final_hash,
+					final_data,
+					dest_encryption.is_encrypted(),
+					None
+				),
+				// Thing 2: we need to insert the block in the version
+				garage.version_table.insert(&dest_version),
+				// Thing 3: we need to add a block reference
+				garage.block_ref_table.insert(&block_ref),
+				// Thing 4: we need to read the next block
+				defragmenter.next(),
+			)?;
+			next
+		} else {
+			blocks_to_dup.blocks.put(version_block_key, version_block);
+			defragmenter.next().await?
 		};
-
-		let (_, _, _, next) = futures::try_join!(
-			// Thing 1: if the block is not exactly a block that existed before,
-			// we need to insert that data as a new block.
-			async {
-				if let Some(final_data) = data_to_upload {
-					garage
-						.block_manager
-						.rpc_put_block(final_hash, final_data, dest_encryption.is_encrypted(), None)
-						.await
-				} else {
-					Ok(())
-				}
-			},
-			// Thing 2: we need to insert the block in the version
-			garage.version_table.insert(&dest_version),
-			// Thing 3: we need to add a block reference
-			garage.block_ref_table.insert(&block_ref),
-			// Thing 4: we need to read the next block
-			defragmenter.next(),
-		)?;
 		next_block = next;
 	}

 	assert_eq!(current_offset, source_range.length);

+	// Put the duplicated blocks into the version & block_refs tables
+	let block_refs_to_put = blocks_to_dup
+		.blocks
+		.items()
+		.iter()
+		.map(|b| BlockRef {
+			block: b.1.hash,
+			version: dest_version_id,
+			deleted: false.into(),
+		})
+		.collect::<Vec<_>>();
+	futures::try_join!(
+		garage.version_table.insert(&blocks_to_dup),
+		garage.block_ref_table.insert_many(&block_refs_to_put[..]),
+	)?;
+
 	let checksums = checksummer.finalize();
 	let etag = dest_encryption.etag_from_md5(&checksums.md5);
-	let checksum = checksums.extract(dest_object_checksum_algorithm);
+	let checksum = checksums.extract(dest_object_checksum_algorithm.map(|(algo, _)| algo));

 	// Put the part's ETag in the Versiontable
 	dest_mpu.parts.put(
@@ -29,7 +29,7 @@ pub async fn handle_get_cors(ctx: ReqCtx) -> Result<Response<ResBody>, Error> {
 			.body(string_body(xml))?)
 	} else {
 		Ok(Response::builder()
-			.status(StatusCode::NO_CONTENT)
+			.status(StatusCode::NOT_FOUND)
 			.body(empty_body())?)
 	}
 }
@@ -176,6 +176,7 @@ impl ApiError for Error {
 		use hyper::header;

 		header_map.append(header::CONTENT_TYPE, "application/xml".parse().unwrap());
+		header_map.append(header::ACCESS_CONTROL_ALLOW_ORIGIN, "*".parse().unwrap());

 		#[allow(clippy::single_match)]
 		match self {
@@ -19,12 +19,13 @@ use garage_net::stream::ByteStream;
 use garage_rpc::rpc_helper::OrderTag;
 use garage_table::EmptyKey;
 use garage_util::data::*;
-use garage_util::error::OkOrMessage;
+use garage_util::error::{Error as UtilError, OkOrMessage};

 use garage_model::garage::Garage;
 use garage_model::s3::object_table::*;
 use garage_model::s3::version_table::*;

+use garage_api_common::common_error::CommonError;
 use garage_api_common::helpers::*;
 use garage_api_common::signature::checksum::{add_checksum_response_headers, X_AMZ_CHECKSUM_MODE};

@@ -219,6 +220,7 @@ pub async fn handle_head_without_ctx(
 					.get(&object_version.uuid, &EmptyKey)
 					.await?
 					.ok_or(Error::NoSuchKey)?;
+				check_version_not_deleted(&version)?;

 				let (part_offset, part_end) =
 					calculate_part_bounds(&version, pn).ok_or(Error::InvalidPart)?;
@@ -373,6 +375,21 @@ pub async fn handle_get_without_ctx(
 	}
 }

+pub(crate) fn check_version_not_deleted(version: &Version) -> Result<(), Error> {
+	if version.deleted.get() {
+		// the version was deleted between when the object_table was consulted
+		// and now, this could mean the object was deleted, or overriden.
+		// Rather than say the key doesn't exist, return a transient error
+		// to signal the client to try again.
+		return Err(CommonError::InternalError(UtilError::Message(
+			"conflict/inconsistency between object and version state, version is deleted"
+				.to_string(),
+		))
+		.into());
+	}
+	Ok(())
+}
+
 async fn handle_get_full(
 	garage: Arc<Garage>,
 	version: &ObjectVersion,
@@ -439,6 +456,7 @@ pub fn full_object_byte_stream(
 						.ok_or_message("channel closed")?;

 					let version = version_fut.await.unwrap()?.ok_or(Error::NoSuchKey)?;
+					check_version_not_deleted(&version)?;
 					for (i, (_, vb)) in version.blocks.items().iter().enumerate().skip(1) {
 						let stream_block_i = encryption
 							.get_block(&garage, &vb.hash, Some(order_stream.order(i as u64)))
@@ -454,6 +472,14 @@ pub fn full_object_byte_stream(
 				{
 					Ok(()) => (),
 					Err(e) => {
+						// TODO i think this is a bad idea, we should log
+						// an error and stop there. If the error happens to
+						// be exactly the size of what hasn't been streamed
+						// yet, the client will see the request as a
+						// success
+						// instead truncating the output notify the client
+						// something happened with their download, so that
+						// they can retry it
 						let _ = tx.send(error_stream_item(e)).await;
 					}
 				}
@@ -505,7 +531,7 @@ async fn handle_get_range(
 				.get(&version.uuid, &EmptyKey)
 				.await?
 				.ok_or(Error::NoSuchKey)?;
-
+			check_version_not_deleted(&version)?;
 			let body =
 				body_from_blocks_range(garage, encryption, version.blocks.items(), begin, end);
 			Ok(resp_builder.body(body)?)
@@ -556,6 +582,8 @@ async fn handle_get_part(
 				.await?
 				.ok_or(Error::NoSuchKey)?;

+			check_version_not_deleted(&version)?;
+
 			let (begin, end) =
 				calculate_part_bounds(&version, part_number).ok_or(Error::InvalidPart)?;

@@ -27,7 +27,7 @@ pub async fn handle_get_lifecycle(ctx: ReqCtx) -> Result<Response<ResBody>, Erro
 			.body(string_body(xml))?)
 	} else {
 		Ok(Response::builder()
-			.status(StatusCode::NO_CONTENT)
+			.status(StatusCode::NOT_FOUND)
 			.body(empty_body())?)
 	}
 }
@@ -1002,6 +1002,7 @@ mod tests {
 					inner: ObjectVersionMetaInner {
 						headers: vec![],
 						checksum: None,
+						checksum_type: None,
 					},
 				},
 				checksum_algorithm: None,
@@ -1,14 +1,12 @@
 use std::collections::HashMap;
-use std::convert::{TryFrom, TryInto};
-use std::hash::Hasher;
+use std::convert::TryInto;
 use std::sync::Arc;

 use base64::prelude::*;
-use crc32c::Crc32cHasher as Crc32c;
-use crc32fast::Hasher as Crc32;
-use crc64fast_nvme::Digest as Crc64Nvme;
+use crc_fast::{CrcAlgorithm, Digest as CrcDigest};
 use futures::prelude::*;
-use hyper::{Request, Response};
+use http::StatusCode;
+use hyper::{header::HeaderValue, HeaderMap, Request, Response};
 use md5::{Digest, Md5};
 use sha1::Sha1;
 use sha2::Sha256;
@@ -54,6 +52,7 @@ pub async fn handle_create_multipart_upload(
 	let meta = ObjectVersionMetaInner {
 		headers,
 		checksum: None,
+		checksum_type: None,
 	};

 	// Determine whether object should be encrypted, and if so the key
@@ -68,7 +67,10 @@ pub async fn handle_create_multipart_upload(
 	)?;
 	let object_encryption = encryption.encrypt_meta(meta)?;

-	let checksum_algorithm = request_checksum_algorithm(req.headers())?;
+	let checksum_algorithm = request_checksum_algorithm_and_type(
+		req.headers(),
+		request_checksum_algorithm(req.headers())?,
+	)?;

 	// Create object in object table
 	let object_version = ObjectVersion {
@@ -227,7 +229,7 @@ pub async fn handle_put_part(
 		MpuPart {
 			version: version_uuid,
 			etag: Some(etag.clone()),
-			checksum: checksums.extract(checksum_algorithm),
+			checksum: checksums.extract(checksum_algorithm.map(|(algo, _)| algo)),
 			size: Some(total_size),
 		},
 	);
@@ -289,12 +291,23 @@ pub async fn handle_complete_multipart_upload(
 	let (req_head, req_body) = req.into_parts();

 	let expected_checksum = request_checksum_value(&req_head.headers)?;
+	let req_checksum_algorithm = request_checksum_algorithm_and_type(
+		&req_head.headers,
+		expected_checksum.map(|x| x.algorithm()),
+	)?;
+	debug!(
+		"CompleteMultipartUpload expected checksum: {:?}, request checksum type: {:?}",
+		expected_checksum, req_checksum_algorithm
+	);

 	let body = req_body.collect().await?;

 	let body_xml = roxmltree::Document::parse(std::str::from_utf8(&body)?)?;
-	let body_list_of_parts = parse_complete_multipart_upload_body(&body_xml)
-		.ok_or_bad_request("Invalid CompleteMultipartUpload XML")?;
+	let body_list_of_parts =
+		parse_complete_multipart_upload_body(&body_xml).ok_or_bad_request(format!(
+			"Invalid CompleteMultipartUpload XML:\n{}",
+			String::from_utf8_lossy(&body)
+		))?;
 	debug!(
 		"CompleteMultipartUpload list of parts: {:?}",
 		body_list_of_parts
@@ -319,6 +332,17 @@ pub async fn handle_complete_multipart_upload(
 		} => (encryption, checksum_algorithm),
 		_ => unreachable!(),
 	};
+	debug!(
+		"CompleteMultipartUpload object checksum_algorithm: {:?}",
+		checksum_algorithm
+	);
+	if req_checksum_algorithm.is_some() && req_checksum_algorithm != checksum_algorithm {
+		return Err(Error::InvalidDigest(format!(
+            "checksum algorithm {:?} does not correspond to algorithm specified in CreateMultipartUpload {:?}",
+            req_checksum_algorithm,
+            checksum_algorithm
+        )));
+	}

 	// Check that part numbers are an increasing sequence.
 	// (it doesn't need to start at 1 nor to be a continuous sequence,
@@ -344,8 +368,7 @@ pub async fn handle_complete_multipart_upload(
 	for req_part in body_list_of_parts.iter() {
 		match have_parts.get(&req_part.part_number) {
 			Some(part) if part.etag.as_ref() == Some(&req_part.etag) && part.size.is_some() => {
-				// alternative version: if req_part.checksum.is_some() && part.checksum != req_part.checksum {
-				if part.checksum != req_part.checksum {
+				if req_part.checksum.is_some() && part.checksum != req_part.checksum {
 					return Err(Error::InvalidDigest(format!(
 						"Invalid checksum for part {}: in request = {:?}, uploaded part = {:?}",
 						req_part.part_number, req_part.checksum, part.checksum
@@ -404,7 +427,11 @@ pub async fn handle_complete_multipart_upload(
 	// https://teppen.io/2018/06/23/aws_s3_etags/
 	let mut checksummer = MultipartChecksummer::init(checksum_algorithm);
 	for part in parts.iter() {
-		checksummer.update(part.etag.as_ref().unwrap(), part.checksum)?;
+		checksummer.update(
+			part.etag.as_ref().unwrap(),
+			part.checksum,
+			part.size.unwrap(),
+		)?;
 	}
 	let (checksum_md5, checksum_extra) = checksummer.finalize();

@@ -440,6 +467,7 @@ pub async fn handle_complete_multipart_upload(
 			let new_meta = ObjectVersionMetaInner {
 				headers: meta.into_owned().headers,
 				checksum: checksum_extra,
+				checksum_type: checksum_algorithm.map(|(_, ty)| ty),
 			};
 			encryption.encrypt_meta(new_meta)?
 		}
@@ -494,6 +522,11 @@ pub async fn handle_complete_multipart_upload(
 			Some(ChecksumValue::Sha256(x)) => Some(s3_xml::Value(BASE64_STANDARD.encode(&x))),
 			_ => None,
 		},
+		checksum_type: match checksum_algorithm {
+			Some((_, ChecksumType::Composite)) => Some(s3_xml::Value(COMPOSITE.into())),
+			Some((_, ChecksumType::FullObject)) => Some(s3_xml::Value(FULL_OBJECT.into())),
+			None => None,
+		},
 	};
 	let xml = s3_xml::to_xml_with_header(&result)?;

@@ -519,7 +552,9 @@ pub async fn handle_abort_multipart_upload(
 	let final_object = Object::new(*bucket_id, key.to_string(), vec![object_version]);
 	garage.object_table.insert(&final_object).await?;

-	Ok(Response::new(empty_body()))
+	Ok(Response::builder()
+		.status(StatusCode::NO_CONTENT)
+		.body(empty_body())?)
 }

 // ======== helpers ============
@@ -571,6 +606,32 @@ struct CompleteMultipartUploadPart {
 	checksum: Option<ChecksumValue>,
 }

+macro_rules! extract_checksum_from {
+	($node:ident { $($name:expr => $variant:ident),* $(,)? }) => {
+		if false { None }
+		$(
+			else if let Some(node) = $node.children().find(|e| e.has_tag_name($name)) {
+				match node.last_child().map(|x| x.text()) {
+					// Child is text but empty post-trim, ignore it.
+					Some(Some(text)) if text.trim().is_empty() => None,
+
+					// Child is non-empty text, parse it.
+					Some(Some(text)) => Some(ChecksumValue::$variant(
+						BASE64_STANDARD.decode(text).ok()?[..].try_into().ok()?
+					)),
+
+					// Child is not text, reject it.
+					Some(None) => return None,
+
+					// No child, ignore it.
+					None => None,
+				}
+			}
+		)*
+		else { None }
+	}
+}
+
 fn parse_complete_multipart_upload_body(
 	xml: &roxmltree::Document,
 ) -> Option<Vec<CompleteMultipartUploadPart>> {
@@ -594,46 +655,15 @@ fn parse_complete_multipart_upload_body(
 				.children()
 				.find(|e| e.has_tag_name("PartNumber"))?
 				.text()?;
-			let checksum = if let Some(crc32) =
-				item.children().find(|e| e.has_tag_name("ChecksumCRC32"))
-			{
-				Some(ChecksumValue::Crc32(
-					BASE64_STANDARD.decode(crc32.text()?).ok()?[..]
-						.try_into()
-						.ok()?,
-				))
-			} else if let Some(crc32c) = item.children().find(|e| e.has_tag_name("ChecksumCRC32C"))
-			{
-				Some(ChecksumValue::Crc32c(
-					BASE64_STANDARD.decode(crc32c.text()?).ok()?[..]
-						.try_into()
-						.ok()?,
-				))
-			} else if let Some(crc64nvme) = item
-				.children()
-				.find(|e| e.has_tag_name("ChecksumCRC64NVME"))
-			{
-				Some(ChecksumValue::Crc64Nvme(
-					BASE64_STANDARD.decode(crc64nvme.text()?).ok()?[..]
-						.try_into()
-						.ok()?,
-				))
-			} else if let Some(sha1) = item.children().find(|e| e.has_tag_name("ChecksumSHA1")) {
-				Some(ChecksumValue::Sha1(
-					BASE64_STANDARD.decode(sha1.text()?).ok()?[..]
-						.try_into()
-						.ok()?,
-				))
-			} else if let Some(sha256) = item.children().find(|e| e.has_tag_name("ChecksumSHA256"))
-			{
-				Some(ChecksumValue::Sha256(
-					BASE64_STANDARD.decode(sha256.text()?).ok()?[..]
-						.try_into()
-						.ok()?,
-				))
-			} else {
-				None
-			};
+
+			let checksum = extract_checksum_from!(item {
+				"ChecksumCRC32" => Crc32,
+				"ChecksumCRC32C" => Crc32c,
+				"ChecksumCRC64NVME" => Crc64Nvme,
+				"ChecksumSHA1" => Sha1,
+				"ChecksumSHA256" => Sha256,
+			});
+
 			parts.push(CompleteMultipartUploadPart {
 				etag: etag.trim_matches('"').to_string(),
 				part_number: part_number.parse().ok()?,
@@ -649,40 +679,52 @@ fn parse_complete_multipart_upload_body(

 // ====== checksummer ====

+pub fn request_checksum_algorithm_and_type(
+	headers: &HeaderMap<HeaderValue>,
+	algo: Option<ChecksumAlgorithm>,
+) -> Result<Option<(ChecksumAlgorithm, ChecksumType)>, Error> {
+	match (headers.get(X_AMZ_CHECKSUM_TYPE), algo) {
+		(None, None) => Ok(None),
+		(None, Some(algo)) => {
+			let ty = match algo {
+				ChecksumAlgorithm::Crc64Nvme => ChecksumType::FullObject,
+				_ => ChecksumType::Composite,
+			};
+			Ok(Some((algo, ty)))
+		}
+		(Some(_), None) => Err(Error::bad_request(
+			"Cannot specify x-amz-checksum-type when no checksum algorithm is in use.",
+		)),
+		(Some(x), Some(algo)) => {
+			let checksum_type = match x.as_bytes() {
+				x if x == COMPOSITE.as_bytes() => ChecksumType::Composite,
+				x if x == FULL_OBJECT.as_bytes() => ChecksumType::FullObject,
+				_ => return Err(Error::bad_request("Invalid x-amz-checksum-type value")),
+			};
+			match (checksum_type, algo) {
+				(ChecksumType::Composite, ChecksumAlgorithm::Crc64Nvme)
+				| (ChecksumType::FullObject, ChecksumAlgorithm::Sha1)
+				| (ChecksumType::FullObject, ChecksumAlgorithm::Sha256) => Err(Error::bad_request(format!(
+					"checksum type {:?} is not supported for algorithm {:?}",
+					checksum_type, algo
+				))),
+				(ty, algo) => Ok(Some((algo, ty))),
+			}
+		}
+	}
+}
+
 #[derive(Default)]
 pub(crate) struct MultipartChecksummer {
 	pub md5: Md5,
 	pub extra: Option<MultipartExtraChecksummer>,
 }

-pub(crate) enum MultipartExtraChecksummer {
-	Crc32(Crc32),
-	Crc32c(Crc32c),
-	Crc64Nvme(Crc64Nvme),
-	Sha1(Sha1),
-	Sha256(Sha256),
-}
-
 impl MultipartChecksummer {
-	pub(crate) fn init(algo: Option<ChecksumAlgorithm>) -> Self {
+	pub(crate) fn init(algo: Option<(ChecksumAlgorithm, ChecksumType)>) -> Self {
 		Self {
 			md5: Md5::new(),
-			extra: match algo {
-				None => None,
-				Some(ChecksumAlgorithm::Crc32) => {
-					Some(MultipartExtraChecksummer::Crc32(Crc32::new()))
-				}
-				Some(ChecksumAlgorithm::Crc32c) => {
-					Some(MultipartExtraChecksummer::Crc32c(Crc32c::default()))
-				}
-				Some(ChecksumAlgorithm::Crc64Nvme) => {
-					Some(MultipartExtraChecksummer::Crc64Nvme(Crc64Nvme::default()))
-				}
-				Some(ChecksumAlgorithm::Sha1) => Some(MultipartExtraChecksummer::Sha1(Sha1::new())),
-				Some(ChecksumAlgorithm::Sha256) => {
-					Some(MultipartExtraChecksummer::Sha256(Sha256::new()))
-				}
-			},
+			extra: algo.map(|(algo, cktype)| MultipartExtraChecksummer::init(algo, cktype)),
 		}
 	}

@@ -690,68 +732,130 @@ impl MultipartChecksummer {
 		&mut self,
 		etag: &str,
 		checksum: Option<ChecksumValue>,
+		part_len: u64,
 	) -> Result<(), Error> {
 		self.md5
 			.update(&hex::decode(&etag).ok_or_message("invalid etag hex")?);
-		match (&mut self.extra, checksum) {
-			(None, _) => (),
-			(
-				Some(MultipartExtraChecksummer::Crc32(ref mut crc32)),
-				Some(ChecksumValue::Crc32(x)),
-			) => {
-				crc32.update(&x);
-			}
-			(
-				Some(MultipartExtraChecksummer::Crc32c(ref mut crc32c)),
-				Some(ChecksumValue::Crc32c(x)),
-			) => {
-				crc32c.write(&x);
-			}
-			(
-				Some(MultipartExtraChecksummer::Crc64Nvme(ref mut crc64nvme)),
-				Some(ChecksumValue::Crc64Nvme(x)),
-			) => {
-				crc64nvme.write(&x);
-			}
-			(Some(MultipartExtraChecksummer::Sha1(ref mut sha1)), Some(ChecksumValue::Sha1(x))) => {
-				sha1.update(&x);
-			}
-			(
-				Some(MultipartExtraChecksummer::Sha256(ref mut sha256)),
-				Some(ChecksumValue::Sha256(x)),
-			) => {
-				sha256.update(&x);
-			}
-			(Some(_), b) => {
-				return Err(Error::internal_error(format!(
-					"part checksum was not computed correctly, got: {:?}",
-					b
-				)))
-			}
+		if let Some(extra) = &mut self.extra {
+			extra.update(checksum, part_len)?;
 		}
 		Ok(())
 	}

 	pub(crate) fn finalize(self) -> (Md5Checksum, Option<ChecksumValue>) {
 		let md5 = self.md5.finalize()[..].try_into().unwrap();
-		let extra = match self.extra {
-			None => None,
-			Some(MultipartExtraChecksummer::Crc32(crc32)) => {
-				Some(ChecksumValue::Crc32(u32::to_be_bytes(crc32.finalize())))
-			}
-			Some(MultipartExtraChecksummer::Crc32c(crc32c)) => Some(ChecksumValue::Crc32c(
-				u32::to_be_bytes(u32::try_from(crc32c.finish()).unwrap()),
-			)),
-			Some(MultipartExtraChecksummer::Crc64Nvme(crc64nvme)) => Some(
-				ChecksumValue::Crc64Nvme(u64::to_be_bytes(crc64nvme.sum64())),
-			),
-			Some(MultipartExtraChecksummer::Sha1(sha1)) => {
-				Some(ChecksumValue::Sha1(sha1.finalize()[..].try_into().unwrap()))
-			}
-			Some(MultipartExtraChecksummer::Sha256(sha256)) => Some(ChecksumValue::Sha256(
-				sha256.finalize()[..].try_into().unwrap(),
-			)),
-		};
+		let extra = self.extra.map(|c| c.finalize());
 		(md5, extra)
 	}
 }
+
+pub(crate) enum MultipartExtraChecksummer {
+	FullObjectCrc(CrcAlgorithm, Option<u64>),
+	CompositeCrc(ChecksumAlgorithm, CrcDigest),
+	CompositeSha1(Sha1),
+	CompositeSha256(Sha256),
+}
+
+impl MultipartExtraChecksummer {
+	fn init(algo: ChecksumAlgorithm, cktype: ChecksumType) -> Self {
+		match (algo, cktype) {
+			(algo, ChecksumType::FullObject) => {
+				let crc_type = match algo {
+					ChecksumAlgorithm::Crc32 => CrcAlgorithm::Crc32IsoHdlc,
+					ChecksumAlgorithm::Crc32c => CrcAlgorithm::Crc32Iscsi,
+					ChecksumAlgorithm::Crc64Nvme => CrcAlgorithm::Crc64Nvme,
+					_ => unreachable!(),
+				};
+				Self::FullObjectCrc(crc_type, None)
+			}
+			(ChecksumAlgorithm::Crc32, ChecksumType::Composite) => {
+				Self::CompositeCrc(ChecksumAlgorithm::Crc32, new_crc32())
+			}
+			(ChecksumAlgorithm::Crc32c, ChecksumType::Composite) => {
+				Self::CompositeCrc(ChecksumAlgorithm::Crc32c, new_crc32c())
+			}
+			(ChecksumAlgorithm::Sha1, ChecksumType::Composite) => Self::CompositeSha1(Sha1::new()),
+			(ChecksumAlgorithm::Sha256, ChecksumType::Composite) => {
+				Self::CompositeSha256(Sha256::new())
+			}
+			_ => unreachable!(),
+		}
+	}
+
+	fn update(&mut self, checksum: Option<ChecksumValue>, part_len: u64) -> Result<(), Error> {
+		match (self, checksum) {
+			(Self::FullObjectCrc(crc_algo, crc_value), Some(ck)) => {
+				let ck_u64 = match ck {
+					ChecksumValue::Crc32(x) => u32::from_be_bytes(x) as u64,
+					ChecksumValue::Crc32c(x) => u32::from_be_bytes(x) as u64,
+					ChecksumValue::Crc64Nvme(x) => u64::from_be_bytes(x),
+					_ => {
+						return Err(Error::internal_error(format!(
+							"part checksum was not computed correctly, got: {:?}",
+							ck
+						)))
+					}
+				};
+				*crc_value = match *crc_value {
+					None => Some(ck_u64),
+					Some(prev) => Some(crc_fast::checksum_combine(
+						*crc_algo, prev, ck_u64, part_len,
+					)),
+				};
+			}
+			(Self::CompositeCrc(_, digest), Some(ck)) => match ck {
+				ChecksumValue::Crc32(x) => digest.update(&x),
+				ChecksumValue::Crc32c(x) => digest.update(&x),
+				ChecksumValue::Crc64Nvme(x) => digest.update(&x),
+				_ => {
+					return Err(Error::internal_error(format!(
+						"part checksum was not computed correctly, got: {:?}",
+						ck
+					)))
+				}
+			},
+			(Self::CompositeSha1(sha1), Some(ChecksumValue::Sha1(x))) => {
+				sha1.update(&x);
+			}
+			(Self::CompositeSha256(sha256), Some(ChecksumValue::Sha256(x))) => {
+				sha256.update(&x);
+			}
+			_ => {
+				return Err(Error::internal_error(format!(
+					"part checksum was not computed correctly, got: {:?}",
+					checksum
+				)))
+			}
+		}
+		Ok(())
+	}
+	fn finalize(self) -> ChecksumValue {
+		match self {
+			Self::FullObjectCrc(algo, value) => match (algo, value) {
+				(CrcAlgorithm::Crc32IsoHdlc, Some(v)) => {
+					ChecksumValue::Crc32(u32::to_be_bytes(v as u32))
+				}
+				(CrcAlgorithm::Crc32Iscsi, Some(v)) => {
+					ChecksumValue::Crc32c(u32::to_be_bytes(v as u32))
+				}
+				(CrcAlgorithm::Crc64Nvme, Some(v)) => ChecksumValue::Crc64Nvme(u64::to_be_bytes(v)),
+				_ => unreachable!(),
+			},
+			Self::CompositeCrc(algo, crc) => match algo {
+				ChecksumAlgorithm::Crc32 => {
+					ChecksumValue::Crc32(u32::to_be_bytes(crc.finalize() as u32))
+				}
+				ChecksumAlgorithm::Crc32c => {
+					ChecksumValue::Crc32c(u32::to_be_bytes(crc.finalize() as u32))
+				}
+				_ => unreachable!(),
+			},
+			Self::CompositeSha1(sha1) => {
+				ChecksumValue::Sha1(sha1.finalize()[..].try_into().unwrap())
+			}
+			Self::CompositeSha256(sha256) => {
+				ChecksumValue::Sha256(sha256.finalize()[..].try_into().unwrap())
+			}
+		}
+	}
+}
@@ -233,6 +233,7 @@ pub async fn handle_post_object(
 	let meta = ObjectVersionMetaInner {
 		headers,
 		checksum: expected_checksums.extra,
+		checksum_type: expected_checksums.extra.map(|_| ChecksumType::FullObject),
 	};

 	let encryption = EncryptionParams::new_from_headers(
@@ -261,7 +262,7 @@ pub async fn handle_post_object(
 		encryption,
 		StreamLimiter::new(stream, conditions.content_length),
 		&key,
-		ChecksumMode::Verify(&expected_checksums),
+		ChecksumMode::Verify(expected_checksums),
 	)
 	.await?;

@@ -48,8 +48,8 @@ pub(crate) struct SaveStreamResult {
 	pub(crate) etag: String,
 }

-pub(crate) enum ChecksumMode<'a> {
-	Verify(&'a ExpectedChecksums),
+pub(crate) enum ChecksumMode {
+	Verify(ExpectedChecksums),
 	VerifyFrom {
 		checksummer: StreamingChecksumReceiver,
 		trailer_algo: Option<ChecksumAlgorithm>,
@@ -83,6 +83,7 @@ pub async fn handle_put(
 	let meta = ObjectVersionMetaInner {
 		headers,
 		checksum: expected_checksums.extra,
+		checksum_type: expected_checksums.extra.map(|_| ChecksumType::FullObject),
 	};

 	// Determine whether object should be encrypted, and if so the key
@@ -139,7 +140,7 @@ pub(crate) async fn save_stream<S: Stream<Item = Result<Bytes, Error>> + Unpin>(
 	encryption: EncryptionParams,
 	body: S,
 	key: &String,
-	checksum_mode: ChecksumMode<'_>,
+	checksum_mode: ChecksumMode,
 ) -> Result<SaveStreamResult, Error> {
 	let ReqCtx {
 		garage, bucket_id, ..
@@ -13,6 +13,10 @@ pub fn xmlns_tag<S: Serializer>(_v: &(), s: S) -> Result<S::Ok, S::Error> {
 	s.serialize_str("http://s3.amazonaws.com/doc/2006-03-01/")
 }

+pub fn xmlns_xsi_tag<S: Serializer>(_v: &(), s: S) -> Result<S::Ok, S::Error> {
+	s.serialize_str("http://www.w3.org/2001/XMLSchema-instance")
+}
+
 #[derive(Debug, Serialize, Deserialize, PartialEq, Eq, PartialOrd, Ord)]
 pub struct Value(#[serde(rename = "$value")] pub String);

@@ -141,6 +145,8 @@ pub struct CompleteMultipartUploadResult {
 	pub checksum_sha1: Option<Value>,
 	#[serde(rename = "ChecksumSHA256")]
 	pub checksum_sha256: Option<Value>,
+	#[serde(rename = "ChecksumType")]
+	pub checksum_type: Option<Value>,
 }

 #[derive(Debug, Serialize, PartialEq, Eq)]
@@ -323,6 +329,42 @@ pub struct PostObject {
 	pub etag: Value,
 }

+#[derive(Debug, Serialize, PartialEq, Eq)]
+pub struct Grantee {
+	#[serde(rename = "xmlns:xsi", serialize_with = "xmlns_xsi_tag")]
+	pub xmlns_xsi: (),
+	#[serde(rename = "xsi:type")]
+	pub typ: String,
+	#[serde(rename = "DisplayName")]
+	pub display_name: Option<Value>,
+	#[serde(rename = "ID")]
+	pub id: Option<Value>,
+}
+
+#[derive(Debug, Serialize, PartialEq, Eq)]
+pub struct Grant {
+	#[serde(rename = "Grantee")]
+	pub grantee: Grantee,
+	#[serde(rename = "Permission")]
+	pub permission: Value,
+}
+
+#[derive(Debug, Serialize, PartialEq, Eq)]
+pub struct AccessControlList {
+	#[serde(rename = "Grant")]
+	pub entries: Vec<Grant>,
+}
+
+#[derive(Debug, Serialize, PartialEq, Eq)]
+pub struct AccessControlPolicy {
+	#[serde(serialize_with = "xmlns_tag")]
+	pub xmlns: (),
+	#[serde(rename = "Owner")]
+	pub owner: Option<Owner>,
+	#[serde(rename = "AccessControlList")]
+	pub acl: AccessControlList,
+}
+
 #[cfg(test)]
 mod tests {
 	use super::*;
@@ -431,6 +473,43 @@ mod tests {
 		Ok(())
 	}

+	#[test]
+	fn get_bucket_acl_result() -> Result<(), ApiError> {
+		let grant = Grant {
+			grantee: Grantee {
+				xmlns_xsi: (),
+				typ: "CanonicalUser".to_string(),
+				display_name: Some(Value("owner_name".to_string())),
+				id: Some(Value("qsdfjklm".to_string())),
+			},
+			permission: Value("FULL_CONTROL".to_string()),
+		};
+
+		let get_bucket_acl = AccessControlPolicy {
+			xmlns: (),
+			owner: None,
+			acl: AccessControlList {
+				entries: vec![grant],
+			},
+		};
+		assert_eq!(
+			to_xml_with_header(&get_bucket_acl)?,
+			"<?xml version=\"1.0\" encoding=\"UTF-8\"?>\
+<AccessControlPolicy xmlns=\"http://s3.amazonaws.com/doc/2006-03-01/\">\
+	<AccessControlList>\
+		<Grant>\
+			<Grantee xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:type=\"CanonicalUser\">\
+				<DisplayName>owner_name</DisplayName>\
+				<ID>qsdfjklm</ID>\
+			</Grantee>\
+			<Permission>FULL_CONTROL</Permission>\
+		</Grant>\
+	</AccessControlList>\
+</AccessControlPolicy>"
+		);
+		Ok(())
+	}
+
 	#[test]
 	fn delete_result() -> Result<(), ApiError> {
 		let delete_result = DeleteResult {
@@ -514,6 +593,7 @@ mod tests {

 	#[test]
 	fn complete_multipart_upload_result() -> Result<(), ApiError> {
+		use garage_api_common::signature::checksum::COMPOSITE;
 		let result = CompleteMultipartUploadResult {
 			xmlns: (),
 			location: Some(Value("https://garage.tld/mybucket/a/plop".to_string())),
@@ -525,6 +605,7 @@ mod tests {
 			checksum_crc64nvme: None,
 			checksum_sha1: Some(Value("ZJAnHyG8PeKz9tI8UTcHrJos39A=".into())),
 			checksum_sha256: None,
+			checksum_type: Some(Value(COMPOSITE.into())),
 		};
 		assert_eq!(
 			to_xml_with_header(&result)?,
@@ -535,6 +616,7 @@ mod tests {
 	<Key>a/plop</Key>\
 	<ETag>&quot;3858f62230ac3c915f300c664312c11f-9&quot;</ETag>\
    <ChecksumSHA1>ZJAnHyG8PeKz9tI8UTcHrJos39A=</ChecksumSHA1>\
+    <ChecksumType>COMPOSITE</ChecksumType>\
 </CompleteMultipartUploadResult>"
 		);
 		Ok(())
@@ -1,6 +1,6 @@
 [package]
 name = "garage_block"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -48,6 +48,8 @@ pub const INLINE_THRESHOLD: usize = 3072;
 // to delete the block locally.
 pub(crate) const BLOCK_GC_DELAY: Duration = Duration::from_secs(600);

+const BLOCK_READ_SEMAPHORE_TIMEOUT: Duration = Duration::from_secs(15);
+
 /// RPC messages used to share blocks of data between nodes
 #[derive(Debug, Serialize, Deserialize)]
 pub enum BlockRpc {
@@ -85,6 +87,7 @@ pub struct BlockManager {
 	disable_scrub: bool,

 	mutation_lock: Vec<Mutex<BlockManagerLocked>>,
+	read_semaphore: Semaphore,

 	pub rc: BlockRc,
 	pub resync: BlockResyncManager,
@@ -174,6 +177,8 @@ impl BlockManager {
 				.iter()
 				.map(|_| Mutex::new(BlockManagerLocked()))
 				.collect::<Vec<_>>(),
+
+			read_semaphore: Semaphore::new(config.block_max_concurrent_reads),
 			rc,
 			resync,
 			system,
@@ -285,7 +290,7 @@ impl BlockManager {
 		let who = self
 			.system
 			.rpc_helper()
-			.block_read_nodes_of(hash, self.system.rpc_helper());
+			.block_read_nodes_of(hash, self.system.rpc_helper())?;

 		for node in who.iter() {
 			let node_id = NodeID::from(*node);
@@ -341,12 +346,9 @@ impl BlockManager {
 	/// layout version only: since blocks are immutable, we don't need to
 	/// do complex logic when several layour versions are active at once,
 	/// just move them directly to the new nodes.
-	pub(crate) fn storage_nodes_of(&self, hash: &Hash) -> Vec<Uuid> {
-		self.system
-			.cluster_layout()
-			.current()
-			.nodes_of(hash)
-			.collect()
+	pub(crate) fn storage_nodes_of(&self, hash: &Hash) -> Result<Vec<Uuid>, Error> {
+		let cluster_layout = self.system.cluster_layout();
+		Ok(cluster_layout.current()?.nodes_of(hash).collect())
 	}

 	// ---- Public interface ----
@@ -381,7 +383,7 @@ impl BlockManager {
 		prevent_compression: bool,
 		order_tag: Option<OrderTag>,
 	) -> Result<(), Error> {
-		let who = self.storage_nodes_of(&hash);
+		let who = self.storage_nodes_of(&hash)?;

 		let compression_level = self.compression_level.filter(|_| !prevent_compression);
 		let (header, bytes) = DataBlock::from_buffer(data, compression_level)
@@ -419,8 +421,8 @@ impl BlockManager {
 	}

 	/// Get number of items in the refcount table
-	pub fn rc_len(&self) -> Result<usize, Error> {
-		Ok(self.rc.rc_table.len()?)
+	pub fn rc_approximate_len(&self) -> Result<usize, Error> {
+		Ok(self.rc.rc_table.approximate_len()?)
 	}

 	/// Send command to start/stop/manager scrub worker
@@ -438,7 +440,7 @@ impl BlockManager {

 	/// List all resync errors
 	pub fn list_resync_errors(&self) -> Result<Vec<BlockResyncErrorInfo>, Error> {
-		let mut blocks = Vec::with_capacity(self.resync.errors.len()?);
+		let mut blocks = Vec::with_capacity(self.resync.errors.approximate_len()?);
 		for ent in self.resync.errors.iter()? {
 			let (hash, cnt) = ent?;
 			let cnt = ErrorCounter::decode(&cnt);
@@ -568,9 +570,6 @@ impl BlockManager {
 			match self.find_block(hash).await {
 				Some(p) => self.read_block_from(hash, &p).await,
 				None => {
-					// Not found but maybe we should have had it ??
-					self.resync
-						.put_to_resync(hash, 2 * self.system.rpc_helper().rpc_timeout())?;
 					return Err(Error::Message(format!(
 						"block {:?} not found on node",
 						hash
@@ -592,6 +591,15 @@ impl BlockManager {
 	) -> Result<DataBlock, Error> {
 		let (header, path) = block_path.as_parts_ref();

+		let permit = tokio::select! {
+			sem = self.read_semaphore.acquire() => sem.ok_or_message("acquire read semaphore")?,
+			_ = tokio::time::sleep(BLOCK_READ_SEMAPHORE_TIMEOUT) => {
+				self.metrics.block_read_semaphore_timeouts.add(1);
+				debug!("read block {:?}: read_semaphore acquire timeout", hash);
+				return Err(Error::Message("read block: read_semaphore acquire timeout".into()));
+			}
+		};
+
 		let mut f = fs::File::open(&path).await?;
 		let mut data = vec![];
 		f.read_to_end(&mut data).await?;
@@ -616,6 +624,8 @@ impl BlockManager {
 			return Err(Error::CorruptData(*hash));
 		}

+		drop(permit);
+
 		Ok(data)
 	}

@@ -22,6 +22,7 @@ pub struct BlockManagerMetrics {

 	pub(crate) bytes_read: BoundCounter<u64>,
 	pub(crate) block_read_duration: BoundValueRecorder<f64>,
+	pub(crate) block_read_semaphore_timeouts: BoundCounter<u64>,
 	pub(crate) bytes_written: BoundCounter<u64>,
 	pub(crate) block_write_duration: BoundValueRecorder<f64>,
 	pub(crate) delete_counter: BoundCounter<u64>,
@@ -50,7 +51,7 @@ impl BlockManagerMetrics {
 				.init(),
 			_rc_size: meter
 				.u64_value_observer("block.rc_size", move |observer| {
-					if let Ok(value) = rc_tree.len() {
+					if let Ok(value) = rc_tree.approximate_len() {
 						observer.observe(value as u64, &[])
 					}
 				})
@@ -58,7 +59,7 @@ impl BlockManagerMetrics {
 				.init(),
 			_resync_queue_len: meter
 				.u64_value_observer("block.resync_queue_length", move |observer| {
-					if let Ok(value) = resync_queue.len() {
+					if let Ok(value) = resync_queue.approximate_len() {
 						observer.observe(value as u64, &[]);
 					}
 				})
@@ -68,7 +69,7 @@ impl BlockManagerMetrics {
 				.init(),
 			_resync_errored_blocks: meter
 				.u64_value_observer("block.resync_errored_blocks", move |observer| {
-					if let Ok(value) = resync_errors.len() {
+					if let Ok(value) = resync_errors.approximate_len() {
 						observer.observe(value as u64, &[]);
 					}
 				})
@@ -119,6 +120,11 @@ impl BlockManagerMetrics {
 				.with_description("Duration of block read operations")
 				.init()
 				.bind(&[]),
+			block_read_semaphore_timeouts: meter
+				.u64_counter("block.read_semaphore_timeouts")
+				.with_description("Number of block reads that failed due to semaphore acquire timeout")
+				.init()
+				.bind(&[]),
 			bytes_written: meter
 				.u64_counter("block.bytes_written")
 				.with_description("Number of bytes written to disk")
@@ -104,13 +104,13 @@ impl BlockResyncManager {
 	}

 	/// Get length of resync queue
-	pub fn queue_len(&self) -> Result<usize, Error> {
-		Ok(self.queue.len()?)
+	pub fn queue_approximate_len(&self) -> Result<usize, Error> {
+		Ok(self.queue.approximate_len()?)
 	}

 	/// Get number of blocks that have an error
-	pub fn errors_len(&self) -> Result<usize, Error> {
-		Ok(self.errors.len()?)
+	pub fn errors_approximate_len(&self) -> Result<usize, Error> {
+		Ok(self.errors.approximate_len()?)
 	}

 	/// Clear the error counter for a block and put it in queue immediately
@@ -131,6 +131,14 @@ impl BlockResyncManager {
 		)))
 	}

+	/// Clear the entire resync queue and list of errored blocks
+	/// Corresponds to `garage repair clear-resync-queue`
+	pub fn clear_resync_queue(&self) -> Result<(), Error> {
+		self.queue.clear()?;
+		self.errors.clear()?;
+		Ok(())
+	}
+
 	pub fn register_bg_vars(&self, vars: &mut vars::BgVars) {
 		let notify = self.notify.clone();
 		vars.register_rw(
@@ -375,7 +383,7 @@ impl BlockResyncManager {
 			info!("Resync block {:?}: offloading and deleting", hash);
 			let existing_path = existing_path.unwrap();

-			let mut who = manager.storage_nodes_of(hash);
+			let mut who = manager.storage_nodes_of(hash)?;
 			if who.len() < manager.write_quorum {
 				return Err(Error::Message("Not trying to offload block because we don't have a quorum of nodes to write to".to_string()));
 			}
@@ -458,7 +466,7 @@ impl BlockResyncManager {

 			// First, check whether we are still supposed to store that
 			// block in the latest cluster layout version.
-			let storage_nodes = manager.storage_nodes_of(&hash);
+			let storage_nodes = manager.storage_nodes_of(&hash)?;

 			if !storage_nodes.contains(&manager.system.id) {
 				info!(
@@ -540,9 +548,11 @@ impl Worker for ResyncWorker {
 		}

 		WorkerStatus {
-			queue_length: Some(self.manager.resync.queue_len().unwrap_or(0) as u64),
+			queue_length: Some(self.manager.resync.queue_approximate_len().unwrap_or(0) as u64),
 			tranquility: Some(tranquility),
-			persistent_errors: Some(self.manager.resync.errors_len().unwrap_or(0) as u64),
+			persistent_errors: Some(
+				self.manager.resync.errors_approximate_len().unwrap_or(0) as u64
+			),
 			..Default::default()
 		}
 	}
@@ -1,6 +1,6 @@
 [package]
 name = "garage_db"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -16,10 +16,14 @@ err-derive.workspace = true
 tracing.workspace = true

 heed = { workspace = true, optional = true }
+
 rusqlite = { workspace = true, optional = true, features = ["backup"] }
 r2d2 = { workspace = true, optional = true }
 r2d2_sqlite = { workspace = true, optional = true }

+fjall = { workspace = true, optional = true }
+parking_lot = { workspace = true, optional = true }
+
 [dev-dependencies]
 mktemp.workspace = true

@@ -27,4 +31,5 @@ mktemp.workspace = true
 default = [ "lmdb", "sqlite" ]
 bundled-libs = [ "rusqlite?/bundled" ]
 lmdb = [ "heed" ]
+fjall = [ "dep:fjall", "dep:parking_lot" ]
 sqlite = [ "rusqlite", "r2d2", "r2d2_sqlite" ]
@@ -0,0 +1,453 @@
+use core::ops::Bound;
+
+use std::path::PathBuf;
+use std::sync::Arc;
+
+use parking_lot::{MappedRwLockReadGuard, RwLock, RwLockReadGuard};
+
+use fjall::{
+	PartitionCreateOptions, PersistMode, TransactionalKeyspace, TransactionalPartitionHandle,
+	WriteTransaction,
+};
+
+use crate::{
+	open::{Engine, OpenOpt},
+	Db, Error, IDb, ITx, ITxFn, OnCommit, Result, TxError, TxFnResult, TxOpError, TxOpResult,
+	TxResult, TxValueIter, Value, ValueIter,
+};
+
+pub use fjall;
+
+// --
+
+pub(crate) fn open_db(path: &PathBuf, opt: &OpenOpt) -> Result<Db> {
+	info!("Opening Fjall database at: {}", path.display());
+	if opt.fsync {
+		return Err(Error(
+			"metadata_fsync is not supported with the Fjall database engine".into(),
+		));
+	}
+	let mut config = fjall::Config::new(path);
+	if let Some(block_cache_size) = opt.fjall_block_cache_size {
+		config = config.cache_size(block_cache_size as u64);
+	}
+	let keyspace = config.open_transactional()?;
+	Ok(FjallDb::init(keyspace))
+}
+
+// -- err
+
+impl From<fjall::Error> for Error {
+	fn from(e: fjall::Error) -> Error {
+		Error(format!("fjall: {}", e).into())
+	}
+}
+
+impl From<fjall::LsmError> for Error {
+	fn from(e: fjall::LsmError) -> Error {
+		Error(format!("fjall lsm_tree: {}", e).into())
+	}
+}
+
+impl From<fjall::Error> for TxOpError {
+	fn from(e: fjall::Error) -> TxOpError {
+		TxOpError(e.into())
+	}
+}
+
+// -- db
+
+pub struct FjallDb {
+	keyspace: TransactionalKeyspace,
+	trees: RwLock<Vec<(String, TransactionalPartitionHandle)>>,
+}
+
+type ByteRefRangeBound<'r> = (Bound<&'r [u8]>, Bound<&'r [u8]>);
+
+impl FjallDb {
+	pub fn init(keyspace: TransactionalKeyspace) -> Db {
+		let s = Self {
+			keyspace,
+			trees: RwLock::new(Vec::new()),
+		};
+		Db(Arc::new(s))
+	}
+
+	fn get_tree(
+		&self,
+		i: usize,
+	) -> Result<MappedRwLockReadGuard<'_, TransactionalPartitionHandle>> {
+		RwLockReadGuard::try_map(self.trees.read(), |trees: &Vec<_>| {
+			trees.get(i).map(|tup| &tup.1)
+		})
+		.map_err(|_| Error("invalid tree id".into()))
+	}
+}
+
+impl IDb for FjallDb {
+	fn engine(&self) -> String {
+		"Fjall (EXPERIMENTAL!)".into()
+	}
+
+	fn open_tree(&self, name: &str) -> Result<usize> {
+		let mut trees = self.trees.write();
+		let safe_name = encode_name(name)?;
+		if let Some(i) = trees.iter().position(|(name, _)| *name == safe_name) {
+			Ok(i)
+		} else {
+			let tree = self
+				.keyspace
+				.open_partition(&safe_name, PartitionCreateOptions::default())?;
+			let i = trees.len();
+			trees.push((safe_name, tree));
+			Ok(i)
+		}
+	}
+
+	fn list_trees(&self) -> Result<Vec<String>> {
+		Ok(self
+			.keyspace
+			.list_partitions()
+			.iter()
+			.map(|n| decode_name(&n))
+			.collect::<Result<Vec<_>>>()?)
+	}
+
+	fn snapshot(&self, base_path: &PathBuf) -> Result<()> {
+		std::fs::create_dir_all(base_path)?;
+		let path = Engine::Fjall.db_path(base_path);
+
+		let source_state = self.keyspace.read_tx();
+		let copy_keyspace = fjall::Config::new(path).open()?;
+
+		for partition_name in self.keyspace.list_partitions() {
+			let source_partition = self
+				.keyspace
+				.open_partition(&partition_name, PartitionCreateOptions::default())?;
+			let copy_partition =
+				copy_keyspace.open_partition(&partition_name, PartitionCreateOptions::default())?;
+
+			for entry in source_state.iter(&source_partition) {
+				let (key, value) = entry?;
+				copy_partition.insert(key, value)?;
+			}
+		}
+
+		copy_keyspace.persist(PersistMode::SyncAll)?;
+		Ok(())
+	}
+
+	// ----
+
+	fn get(&self, tree_idx: usize, key: &[u8]) -> Result<Option<Value>> {
+		let tree = self.get_tree(tree_idx)?;
+		let tx = self.keyspace.read_tx();
+		let val = tx.get(&tree, key)?;
+		match val {
+			None => Ok(None),
+			Some(v) => Ok(Some(v.to_vec())),
+		}
+	}
+
+	fn approximate_len(&self, tree_idx: usize) -> Result<usize> {
+		let tree = self.get_tree(tree_idx)?;
+		Ok(tree.approximate_len())
+	}
+	fn is_empty(&self, tree_idx: usize) -> Result<bool> {
+		let tree = self.get_tree(tree_idx)?;
+		let tx = self.keyspace.read_tx();
+		Ok(tx.is_empty(&tree)?)
+	}
+
+	fn insert(&self, tree_idx: usize, key: &[u8], value: &[u8]) -> Result<()> {
+		let tree = self.get_tree(tree_idx)?;
+		let mut tx = self.keyspace.write_tx();
+		tx.insert(&tree, key, value);
+		tx.commit()?;
+		Ok(())
+	}
+
+	fn remove(&self, tree_idx: usize, key: &[u8]) -> Result<()> {
+		let tree = self.get_tree(tree_idx)?;
+		let mut tx = self.keyspace.write_tx();
+		tx.remove(&tree, key);
+		tx.commit()?;
+		Ok(())
+	}
+
+	fn clear(&self, tree_idx: usize) -> Result<()> {
+		let mut trees = self.trees.write();
+
+		if tree_idx >= trees.len() {
+			return Err(Error("invalid tree id".into()));
+		}
+		let (name, tree) = trees.remove(tree_idx);
+
+		self.keyspace.delete_partition(tree)?;
+		let tree = self
+			.keyspace
+			.open_partition(&name, PartitionCreateOptions::default())?;
+		trees.insert(tree_idx, (name, tree));
+
+		Ok(())
+	}
+
+	fn iter(&self, tree_idx: usize) -> Result<ValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?;
+		let tx = self.keyspace.read_tx();
+		Ok(Box::new(tx.iter(&tree).map(iterator_remap)))
+	}
+
+	fn iter_rev(&self, tree_idx: usize) -> Result<ValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?;
+		let tx = self.keyspace.read_tx();
+		Ok(Box::new(tx.iter(&tree).rev().map(iterator_remap)))
+	}
+
+	fn range<'r>(
+		&self,
+		tree_idx: usize,
+		low: Bound<&'r [u8]>,
+		high: Bound<&'r [u8]>,
+	) -> Result<ValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?;
+		let tx = self.keyspace.read_tx();
+		Ok(Box::new(
+			tx.range::<&'r [u8], ByteRefRangeBound>(&tree, (low, high))
+				.map(iterator_remap),
+		))
+	}
+	fn range_rev<'r>(
+		&self,
+		tree_idx: usize,
+		low: Bound<&'r [u8]>,
+		high: Bound<&'r [u8]>,
+	) -> Result<ValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?;
+		let tx = self.keyspace.read_tx();
+		Ok(Box::new(
+			tx.range::<&'r [u8], ByteRefRangeBound>(&tree, (low, high))
+				.rev()
+				.map(iterator_remap),
+		))
+	}
+
+	// ----
+
+	fn transaction(&self, f: &dyn ITxFn) -> TxResult<OnCommit, ()> {
+		let trees = self.trees.read();
+		let mut tx = FjallTx {
+			trees: &trees[..],
+			tx: self.keyspace.write_tx(),
+		};
+
+		let res = f.try_on(&mut tx);
+		match res {
+			TxFnResult::Ok(on_commit) => {
+				tx.tx.commit().map_err(Error::from).map_err(TxError::Db)?;
+				Ok(on_commit)
+			}
+			TxFnResult::Abort => {
+				tx.tx.rollback();
+				Err(TxError::Abort(()))
+			}
+			TxFnResult::DbErr => {
+				tx.tx.rollback();
+				Err(TxError::Db(Error(
+					"(this message will be discarded)".into(),
+				)))
+			}
+		}
+	}
+}
+
+// ----
+
+struct FjallTx<'a> {
+	trees: &'a [(String, TransactionalPartitionHandle)],
+	tx: WriteTransaction<'a>,
+}
+
+impl<'a> FjallTx<'a> {
+	fn get_tree(&self, i: usize) -> TxOpResult<&TransactionalPartitionHandle> {
+		self.trees.get(i).map(|tup| &tup.1).ok_or_else(|| {
+			TxOpError(Error(
+				"invalid tree id (it might have been openned after the transaction started)".into(),
+			))
+		})
+	}
+}
+
+impl<'a> ITx for FjallTx<'a> {
+	fn get(&self, tree_idx: usize, key: &[u8]) -> TxOpResult<Option<Value>> {
+		let tree = self.get_tree(tree_idx)?;
+		match self.tx.get(tree, key)? {
+			Some(v) => Ok(Some(v.to_vec())),
+			None => Ok(None),
+		}
+	}
+	fn len(&self, tree_idx: usize) -> TxOpResult<usize> {
+		let tree = self.get_tree(tree_idx)?;
+		Ok(self.tx.len(tree)? as usize)
+	}
+
+	fn insert(&mut self, tree_idx: usize, key: &[u8], value: &[u8]) -> TxOpResult<()> {
+		let tree = self.get_tree(tree_idx)?.clone();
+		self.tx.insert(&tree, key, value);
+		Ok(())
+	}
+	fn remove(&mut self, tree_idx: usize, key: &[u8]) -> TxOpResult<()> {
+		let tree = self.get_tree(tree_idx)?.clone();
+		self.tx.remove(&tree, key);
+		Ok(())
+	}
+	fn clear(&mut self, _tree_idx: usize) -> TxOpResult<()> {
+		unimplemented!("LSM tree clearing in cross-partition transaction is not supported")
+	}
+
+	fn iter(&self, tree_idx: usize) -> TxOpResult<TxValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?.clone();
+		Ok(Box::new(self.tx.iter(&tree).map(iterator_remap_tx)))
+	}
+	fn iter_rev(&self, tree_idx: usize) -> TxOpResult<TxValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?.clone();
+		Ok(Box::new(self.tx.iter(&tree).rev().map(iterator_remap_tx)))
+	}
+
+	fn range<'r>(
+		&self,
+		tree_idx: usize,
+		low: Bound<&'r [u8]>,
+		high: Bound<&'r [u8]>,
+	) -> TxOpResult<TxValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?;
+		let low = clone_bound(low);
+		let high = clone_bound(high);
+		Ok(Box::new(
+			self.tx
+				.range::<Vec<u8>, ByteVecRangeBounds>(&tree, (low, high))
+				.map(iterator_remap_tx),
+		))
+	}
+	fn range_rev<'r>(
+		&self,
+		tree_idx: usize,
+		low: Bound<&'r [u8]>,
+		high: Bound<&'r [u8]>,
+	) -> TxOpResult<TxValueIter<'_>> {
+		let tree = self.get_tree(tree_idx)?;
+		let low = clone_bound(low);
+		let high = clone_bound(high);
+		Ok(Box::new(
+			self.tx
+				.range::<Vec<u8>, ByteVecRangeBounds>(&tree, (low, high))
+				.rev()
+				.map(iterator_remap_tx),
+		))
+	}
+}
+
+// -- maps fjall's (k, v) to ours
+
+fn iterator_remap(r: fjall::Result<(fjall::Slice, fjall::Slice)>) -> Result<(Value, Value)> {
+	r.map(|(k, v)| (k.to_vec(), v.to_vec()))
+		.map_err(|e| e.into())
+}
+
+fn iterator_remap_tx(r: fjall::Result<(fjall::Slice, fjall::Slice)>) -> TxOpResult<(Value, Value)> {
+	r.map(|(k, v)| (k.to_vec(), v.to_vec()))
+		.map_err(|e| e.into())
+}
+
+// -- utils to deal with Garage's tightness on Bound lifetimes
+
+type ByteVecBound = Bound<Vec<u8>>;
+type ByteVecRangeBounds = (ByteVecBound, ByteVecBound);
+
+fn clone_bound(bound: Bound<&[u8]>) -> ByteVecBound {
+	let value = match bound {
+		Bound::Excluded(v) | Bound::Included(v) => v.to_vec(),
+		Bound::Unbounded => vec![],
+	};
+
+	match bound {
+		Bound::Included(_) => Bound::Included(value),
+		Bound::Excluded(_) => Bound::Excluded(value),
+		Bound::Unbounded => Bound::Unbounded,
+	}
+}
+
+// -- utils to encode table names --
+
+fn encode_name(s: &str) -> Result<String> {
+	let base = 'A' as u32;
+
+	let mut ret = String::with_capacity(s.len() + 10);
+	for c in s.chars() {
+		if c.is_alphanumeric() || c == '_' || c == '-' || c == '#' {
+			ret.push(c);
+		} else if c <= u8::MAX as char {
+			ret.push('$');
+			let c_hi = c as u32 / 16;
+			let c_lo = c as u32 % 16;
+			ret.push(char::from_u32(base + c_hi).unwrap());
+			ret.push(char::from_u32(base + c_lo).unwrap());
+		} else {
+			return Err(Error(
+				format!("table name {} could not be safely encoded", s).into(),
+			));
+		}
+	}
+	Ok(ret)
+}
+
+fn decode_name(s: &str) -> Result<String> {
+	use std::convert::TryFrom;
+
+	let errfn = || Error(format!("encoded table name {} is invalid", s).into());
+	let c_map = |c: char| {
+		let c = c as u32;
+		let base = 'A' as u32;
+		if (base..base + 16).contains(&c) {
+			Some(c - base)
+		} else {
+			None
+		}
+	};
+
+	let mut ret = String::with_capacity(s.len());
+	let mut it = s.chars();
+	while let Some(c) = it.next() {
+		if c == '$' {
+			let c_hi = it.next().and_then(c_map).ok_or_else(errfn)?;
+			let c_lo = it.next().and_then(c_map).ok_or_else(errfn)?;
+			let c_dec = char::try_from(c_hi * 16 + c_lo).map_err(|_| errfn())?;
+			ret.push(c_dec);
+		} else {
+			ret.push(c);
+		}
+	}
+	Ok(ret)
+}
+
+#[cfg(test)]
+mod tests {
+	use super::*;
+
+	#[test]
+	fn test_encdec_name() {
+		for name in [
+			"testname",
+			"test_name",
+			"test name",
+			"test$name",
+			"test:name@help.me$get/this**right",
+		] {
+			let encname = encode_name(name).unwrap();
+			assert!(!encname.contains(' '));
+			assert!(!encname.contains('.'));
+			assert!(!encname.contains('*'));
+			assert_eq!(*name, decode_name(&encname).unwrap());
+		}
+	}
+}
@@ -1,6 +1,8 @@
 #[macro_use]
 extern crate tracing;

+#[cfg(feature = "fjall")]
+pub mod fjall_adapter;
 #[cfg(feature = "lmdb")]
 pub mod lmdb_adapter;
 #[cfg(feature = "sqlite")]
@@ -104,32 +106,44 @@ impl Db {
 			result: Cell::new(None),
 		};
 		let tx_res = self.0.transaction(&f);
-		let ret = f
-			.result
-			.into_inner()
-			.expect("Transaction did not store result");
+		let fn_res = f.result.into_inner();

-		match tx_res {
-			Ok(on_commit) => match ret {
-				Ok(value) => {
-					on_commit.into_iter().for_each(|f| f());
-					Ok(value)
-				}
-				_ => unreachable!(),
-			},
-			Err(TxError::Abort(())) => match ret {
-				Err(TxError::Abort(e)) => Err(TxError::Abort(e)),
-				_ => unreachable!(),
-			},
-			Err(TxError::Db(e2)) => match ret {
-				// Ok was stored -> the error occurred when finalizing
-				// transaction
-				Ok(_) => Err(TxError::Db(e2)),
-				// An error was already stored: that's the one we want to
-				// return
-				Err(TxError::Db(e)) => Err(TxError::Db(e)),
-				_ => unreachable!(),
-			},
+		match (tx_res, fn_res) {
+			(Ok(on_commit), Some(Ok(value))) => {
+				// Transaction succeeded
+				// TxFn stored the value to return to the user in fn_res
+				// tx_res contains the on_commit list of callbacks, run them now
+				on_commit.into_iter().for_each(|f| f());
+				Ok(value)
+			}
+			(Err(TxError::Abort(())), Some(Err(TxError::Abort(e)))) => {
+				// Transaction was aborted by user code
+				// The abort error value is stored in fn_res
+				Err(TxError::Abort(e))
+			}
+			(Err(TxError::Db(_tx_e)), Some(Err(TxError::Db(fn_e)))) => {
+				// Transaction encountered a DB error in user code
+				// The error value encountered is the one in fn_res,
+				// tx_res contains only a dummy error message
+				Err(TxError::Db(fn_e))
+			}
+			(Err(TxError::Db(tx_e)), None) => {
+				// Transaction encounterred a DB error when initializing the transaction,
+				// before user code was called
+				Err(TxError::Db(tx_e))
+			}
+			(Err(TxError::Db(tx_e)), Some(Ok(_))) => {
+				// Transaction encounterred a DB error when commiting the transaction,
+				// after user code was called
+				Err(TxError::Db(tx_e))
+			}
+			(tx_res, fn_res) => {
+				panic!(
+					"unexpected error case: tx_res={:?}, fn_res={:?}",
+					tx_res.map(|_| "..."),
+					fn_res.map(|x| x.map(|_| "...").map_err(|_| "..."))
+				);
+			}
 		}
 	}

@@ -152,7 +166,7 @@ impl Db {
 		let tree_names = other.list_trees()?;
 		for name in tree_names {
 			let tree = self.open_tree(&name)?;
-			if tree.len()? > 0 {
+			if !tree.is_empty()? {
 				return Err(Error(format!("tree {} already contains data", name).into()));
 			}

@@ -194,8 +208,12 @@ impl Tree {
 		self.0.get(self.1, key.as_ref())
 	}
 	#[inline]
-	pub fn len(&self) -> Result<usize> {
-		self.0.len(self.1)
+	pub fn approximate_len(&self) -> Result<usize> {
+		self.0.approximate_len(self.1)
+	}
+	#[inline]
+	pub fn is_empty(&self) -> Result<bool> {
+		self.0.is_empty(self.1)
 	}

 	#[inline]
@@ -333,7 +351,8 @@ pub(crate) trait IDb: Send + Sync {
 	fn snapshot(&self, path: &PathBuf) -> Result<()>;

 	fn get(&self, tree: usize, key: &[u8]) -> Result<Option<Value>>;
-	fn len(&self, tree: usize) -> Result<usize>;
+	fn approximate_len(&self, tree: usize) -> Result<usize>;
+	fn is_empty(&self, tree: usize) -> Result<bool>;

 	fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<()>;
 	fn remove(&self, tree: usize, key: &[u8]) -> Result<()>;
@@ -1,8 +1,8 @@
 use core::ops::Bound;
-use core::ptr::NonNull;

 use std::collections::HashMap;
 use std::convert::TryInto;
+use std::marker::PhantomPinned;
 use std::path::PathBuf;
 use std::pin::Pin;
 use std::sync::{Arc, RwLock};
@@ -11,12 +11,55 @@ use heed::types::ByteSlice;
 use heed::{BytesDecode, Env, RoTxn, RwTxn, UntypedDatabase as Database};

 use crate::{
+	open::{Engine, OpenOpt},
 	Db, Error, IDb, ITx, ITxFn, OnCommit, Result, TxError, TxFnResult, TxOpError, TxOpResult,
 	TxResult, TxValueIter, Value, ValueIter,
 };

 pub use heed;

+// ---- top-level open function
+
+pub(crate) fn open_db(path: &PathBuf, opt: &OpenOpt) -> Result<Db> {
+	info!("Opening LMDB database at: {}", path.display());
+	if let Err(e) = std::fs::create_dir_all(&path) {
+		return Err(Error(
+			format!("Unable to create LMDB data directory: {}", e).into(),
+		));
+	}
+
+	let map_size = match opt.lmdb_map_size {
+		None => recommended_map_size(),
+		Some(v) => v - (v % 4096),
+	};
+
+	let mut env_builder = heed::EnvOpenOptions::new();
+	env_builder.max_dbs(100);
+	env_builder.map_size(map_size);
+	env_builder.max_readers(2048);
+	unsafe {
+		env_builder.flag(heed::flags::Flags::MdbNoRdAhead);
+		env_builder.flag(heed::flags::Flags::MdbNoMetaSync);
+		if !opt.fsync {
+			env_builder.flag(heed::flags::Flags::MdbNoSync);
+		}
+	}
+	match env_builder.open(&path) {
+		Err(heed::Error::Io(e)) if e.kind() == std::io::ErrorKind::OutOfMemory => {
+			return Err(Error(
+				"OutOfMemory error while trying to open LMDB database. This can happen \
+                if your operating system is not allowing you to use sufficient virtual \
+                memory address space. Please check that no limit is set (ulimit -v). \
+                You may also try to set a smaller `lmdb_map_size` configuration parameter. \
+                On 32-bit machines, you should probably switch to another database engine."
+					.into(),
+			))
+		}
+		Err(e) => Err(Error(format!("Cannot open LMDB database: {}", e).into())),
+		Ok(db) => Ok(LmdbDb::init(db)),
+	}
+}
+
 // -- err

 impl From<heed::Error> for Error {
@@ -104,12 +147,11 @@ impl IDb for LmdbDb {
 		Ok(ret2)
 	}

-	fn snapshot(&self, to: &PathBuf) -> Result<()> {
-		std::fs::create_dir_all(to)?;
-		let mut path = to.clone();
-		path.push("data.mdb");
+	fn snapshot(&self, base_path: &PathBuf) -> Result<()> {
+		std::fs::create_dir_all(base_path)?;
+		let path = Engine::Lmdb.db_path(base_path);
 		self.db
-			.copy_to_path(path, heed::CompactionOption::Disabled)?;
+			.copy_to_path(path, heed::CompactionOption::Enabled)?;
 		Ok(())
 	}

@@ -126,11 +168,16 @@ impl IDb for LmdbDb {
 		}
 	}

-	fn len(&self, tree: usize) -> Result<usize> {
+	fn approximate_len(&self, tree: usize) -> Result<usize> {
 		let tree = self.get_tree(tree)?;
 		let tx = self.db.read_txn()?;
 		Ok(tree.len(&tx)?.try_into().unwrap())
 	}
+	fn is_empty(&self, tree: usize) -> Result<bool> {
+		let tree = self.get_tree(tree)?;
+		let tx = self.db.read_txn()?;
+		Ok(tree.is_empty(&tx)?)
+	}

 	fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<()> {
 		let tree = self.get_tree(tree)?;
@@ -159,13 +206,15 @@ impl IDb for LmdbDb {
 	fn iter(&self, tree: usize) -> Result<ValueIter<'_>> {
 		let tree = self.get_tree(tree)?;
 		let tx = self.db.read_txn()?;
-		TxAndIterator::make(tx, |tx| Ok(tree.iter(tx)?))
+		// Safety: the cloture does not store its argument anywhere,
+		unsafe { TxAndIterator::make(tx, |tx| Ok(tree.iter(tx)?)) }
 	}

 	fn iter_rev(&self, tree: usize) -> Result<ValueIter<'_>> {
 		let tree = self.get_tree(tree)?;
 		let tx = self.db.read_txn()?;
-		TxAndIterator::make(tx, |tx| Ok(tree.rev_iter(tx)?))
+		// Safety: the cloture does not store its argument anywhere,
+		unsafe { TxAndIterator::make(tx, |tx| Ok(tree.rev_iter(tx)?)) }
 	}

 	fn range<'r>(
@@ -176,7 +225,8 @@ impl IDb for LmdbDb {
 	) -> Result<ValueIter<'_>> {
 		let tree = self.get_tree(tree)?;
 		let tx = self.db.read_txn()?;
-		TxAndIterator::make(tx, |tx| Ok(tree.range(tx, &(low, high))?))
+		// Safety: the cloture does not store its argument anywhere,
+		unsafe { TxAndIterator::make(tx, |tx| Ok(tree.range(tx, &(low, high))?)) }
 	}
 	fn range_rev<'r>(
 		&self,
@@ -186,7 +236,8 @@ impl IDb for LmdbDb {
 	) -> Result<ValueIter<'_>> {
 		let tree = self.get_tree(tree)?;
 		let tx = self.db.read_txn()?;
-		TxAndIterator::make(tx, |tx| Ok(tree.rev_range(tx, &(low, high))?))
+		// Safety: the cloture does not store its argument anywhere,
+		unsafe { TxAndIterator::make(tx, |tx| Ok(tree.rev_range(tx, &(low, high))?)) }
 	}

 	// ----
@@ -316,28 +367,41 @@ where
 {
 	tx: RoTxn<'a>,
 	iter: Option<I>,
+	_pin: PhantomPinned,
 }

 impl<'a, I> TxAndIterator<'a, I>
 where
 	I: Iterator<Item = IteratorItem<'a>> + 'a,
 {
-	fn make<F>(tx: RoTxn<'a>, iterfun: F) -> Result<ValueIter<'a>>
+	fn iter(self: Pin<&mut Self>) -> &mut Option<I> {
+		// Safety: iter is not structural
+		unsafe { &mut self.get_unchecked_mut().iter }
+	}
+
+	/// Safety: iterfun must not store its argument anywhere but in its result.
+	unsafe fn make<F>(tx: RoTxn<'a>, iterfun: F) -> Result<ValueIter<'a>>
 	where
 		F: FnOnce(&'a RoTxn<'a>) -> Result<I>,
 	{
-		let res = TxAndIterator { tx, iter: None };
+		let res = TxAndIterator {
+			tx,
+			iter: None,
+			_pin: PhantomPinned,
+		};
 		let mut boxed = Box::pin(res);

-		// This unsafe allows us to bypass lifetime checks
-		let tx = unsafe { NonNull::from(&boxed.tx).as_ref() };
-		let iter = iterfun(tx)?;
+		let tx_lifetime_overextended: &'a RoTxn<'a> = {
+			let tx = &boxed.tx;
+			// Safety: Artificially extending the lifetime because
+			// this reference will only be stored and accessed from the
+			// returned ValueIter which guarantees that it is destroyed
+			// before the tx it is pointing  to.
+			unsafe { &*&raw const *tx }
+		};
+		let iter = iterfun(&tx_lifetime_overextended)?;

-		let mut_ref = Pin::as_mut(&mut boxed);
-		// This unsafe allows us to write in a field of the pinned struct
-		unsafe {
-			Pin::get_unchecked_mut(mut_ref).iter = Some(iter);
-		}
+		*boxed.as_mut().iter() = Some(iter);

 		Ok(Box::new(TxAndIteratorPin(boxed)))
 	}
@@ -348,8 +412,10 @@ where
 	I: Iterator<Item = IteratorItem<'a>> + 'a,
 {
 	fn drop(&mut self) {
-		// ensure the iterator is dropped before the RoTxn it references
-		drop(self.iter.take());
+		// Safety: `new_unchecked` is okay because we know this value is never
+		// used again after being dropped.
+		let this = unsafe { Pin::new_unchecked(self) };
+		drop(this.iter().take());
 	}
 }

@@ -365,13 +431,12 @@ where

 	fn next(&mut self) -> Option<Self::Item> {
 		let mut_ref = Pin::as_mut(&mut self.0);
-		// This unsafe allows us to mutably access the iterator field
-		let next = unsafe { Pin::get_unchecked_mut(mut_ref).iter.as_mut()?.next() };
-		match next {
-			None => None,
-			Some(Err(e)) => Some(Err(e.into())),
-			Some(Ok((k, v))) => Some(Ok((k.to_vec(), v.to_vec()))),
-		}
+		let next = mut_ref.iter().as_mut()?.next()?;
+		let res = match next {
+			Err(e) => Err(e.into()),
+			Ok((k, v)) => Ok((k.to_vec(), v.to_vec())),
+		};
+		Some(res)
 	}
 }

@@ -11,6 +11,7 @@ use crate::{Db, Error, Result};
 pub enum Engine {
 	Lmdb,
 	Sqlite,
+	Fjall,
 }

 impl Engine {
@@ -19,8 +20,26 @@ impl Engine {
 		match self {
 			Self::Lmdb => "lmdb",
 			Self::Sqlite => "sqlite",
+			Self::Fjall => "fjall",
 		}
 	}
+
+	/// Return engine-specific DB path from base path
+	pub fn db_path(&self, base_path: &PathBuf) -> PathBuf {
+		let mut ret = base_path.clone();
+		match self {
+			Self::Lmdb => {
+				ret.push("db.lmdb");
+			}
+			Self::Sqlite => {
+				ret.push("db.sqlite");
+			}
+			Self::Fjall => {
+				ret.push("db.fjall");
+			}
+		}
+		ret
+	}
 }

 impl std::fmt::Display for Engine {
@@ -36,10 +55,11 @@ impl std::str::FromStr for Engine {
 		match text {
 			"lmdb" | "heed" => Ok(Self::Lmdb),
 			"sqlite" | "sqlite3" | "rusqlite" => Ok(Self::Sqlite),
+            "fjall" => Ok(Self::Fjall),
 			"sled" => Err(Error("Sled is no longer supported as a database engine. Converting your old metadata db can be done using an older Garage binary (e.g. v0.9.4).".into())),
 			kind => Err(Error(
 				format!(
-					"Invalid DB engine: {} (options are: lmdb, sqlite)",
+					"Invalid DB engine: {} (options are: lmdb, sqlite, fjall)",
 					kind
 				)
 				.into(),
@@ -51,6 +71,7 @@ impl std::str::FromStr for Engine {
 pub struct OpenOpt {
 	pub fsync: bool,
 	pub lmdb_map_size: Option<usize>,
+	pub fjall_block_cache_size: Option<usize>,
 }

 impl Default for OpenOpt {
@@ -58,6 +79,7 @@ impl Default for OpenOpt {
 		Self {
 			fsync: false,
 			lmdb_map_size: None,
+			fjall_block_cache_size: None,
 		}
 	}
 }
@@ -66,53 +88,15 @@ pub fn open_db(path: &PathBuf, engine: Engine, opt: &OpenOpt) -> Result<Db> {
 	match engine {
 		// ---- Sqlite DB ----
 		#[cfg(feature = "sqlite")]
-		Engine::Sqlite => {
-			info!("Opening Sqlite database at: {}", path.display());
-			let manager = r2d2_sqlite::SqliteConnectionManager::file(path);
-			Ok(crate::sqlite_adapter::SqliteDb::new(manager, opt.fsync)?)
-		}
+		Engine::Sqlite => crate::sqlite_adapter::open_db(path, opt),

 		// ---- LMDB DB ----
 		#[cfg(feature = "lmdb")]
-		Engine::Lmdb => {
-			info!("Opening LMDB database at: {}", path.display());
-			if let Err(e) = std::fs::create_dir_all(&path) {
-				return Err(Error(
-					format!("Unable to create LMDB data directory: {}", e).into(),
-				));
-			}
+		Engine::Lmdb => crate::lmdb_adapter::open_db(path, opt),

-			let map_size = match opt.lmdb_map_size {
-				None => crate::lmdb_adapter::recommended_map_size(),
-				Some(v) => v - (v % 4096),
-			};
-
-			let mut env_builder = heed::EnvOpenOptions::new();
-			env_builder.max_dbs(100);
-			env_builder.map_size(map_size);
-			env_builder.max_readers(2048);
-			unsafe {
-				env_builder.flag(crate::lmdb_adapter::heed::flags::Flags::MdbNoRdAhead);
-				env_builder.flag(crate::lmdb_adapter::heed::flags::Flags::MdbNoMetaSync);
-				if !opt.fsync {
-					env_builder.flag(heed::flags::Flags::MdbNoSync);
-				}
-			}
-			match env_builder.open(&path) {
-				Err(heed::Error::Io(e)) if e.kind() == std::io::ErrorKind::OutOfMemory => {
-					return Err(Error(
-						"OutOfMemory error while trying to open LMDB database. This can happen \
-                        if your operating system is not allowing you to use sufficient virtual \
-                        memory address space. Please check that no limit is set (ulimit -v). \
-                        You may also try to set a smaller `lmdb_map_size` configuration parameter. \
-                        On 32-bit machines, you should probably switch to another database engine."
-							.into(),
-					))
-				}
-				Err(e) => Err(Error(format!("Cannot open LMDB database: {}", e).into())),
-				Ok(db) => Ok(crate::lmdb_adapter::LmdbDb::init(db)),
-			}
-		}
+		// ---- Fjall DB ----
+		#[cfg(feature = "fjall")]
+		Engine::Fjall => crate::fjall_adapter::open_db(path, opt),

 		// Pattern is unreachable when all supported DB engines are compiled into binary. The allow
 		// attribute is added so that we won't have to change this match in case stop building
@@ -11,12 +11,23 @@ use r2d2_sqlite::SqliteConnectionManager;
 use rusqlite::{params, Rows, Statement, Transaction};

 use crate::{
+	open::{Engine, OpenOpt},
 	Db, Error, IDb, ITx, ITxFn, OnCommit, Result, TxError, TxFnResult, TxOpError, TxOpResult,
 	TxResult, TxValueIter, Value, ValueIter,
 };

 pub use rusqlite;

+// ---- top-level open function
+
+pub(crate) fn open_db(path: &PathBuf, opt: &OpenOpt) -> Result<Db> {
+	info!("Opening Sqlite database at: {}", path.display());
+	let manager = r2d2_sqlite::SqliteConnectionManager::file(path);
+	Ok(SqliteDb::new(manager, opt.fsync)?)
+}
+
+// ----
+
 type Connection = r2d2::PooledConnection<SqliteConnectionManager>;

 // --- err
@@ -139,17 +150,18 @@ impl IDb for SqliteDb {
 		Ok(trees)
 	}

-	fn snapshot(&self, to: &PathBuf) -> Result<()> {
-		fn progress(p: rusqlite::backup::Progress) {
-			let percent = (p.pagecount - p.remaining) * 100 / p.pagecount;
-			info!("Sqlite snapshot progress: {}%", percent);
-		}
-		std::fs::create_dir_all(to)?;
-		let mut path = to.clone();
-		path.push("db.sqlite");
-		self.db
-			.get()?
-			.backup(rusqlite::DatabaseName::Main, path, Some(progress))?;
+	fn snapshot(&self, base_path: &PathBuf) -> Result<()> {
+		std::fs::create_dir_all(base_path)?;
+		let path = Engine::Sqlite
+			.db_path(&base_path)
+			.into_os_string()
+			.into_string()
+			.map_err(|_| Error("invalid sqlite path string".into()))?;
+
+		info!("Start sqlite VACUUM INTO `{}`", path);
+		self.db.get()?.execute("VACUUM INTO ?1", params![path])?;
+		info!("Finished sqlite VACUUM INTO `{}`", path);
+
 		Ok(())
 	}

@@ -160,7 +172,7 @@ impl IDb for SqliteDb {
 		self.internal_get(&self.db.get()?, &tree, key)
 	}

-	fn len(&self, tree: usize) -> Result<usize> {
+	fn approximate_len(&self, tree: usize) -> Result<usize> {
 		let tree = self.get_tree(tree)?;
 		let db = self.db.get()?;

@@ -172,6 +184,10 @@ impl IDb for SqliteDb {
 		}
 	}

+	fn is_empty(&self, tree: usize) -> Result<bool> {
+		Ok(self.approximate_len(tree)? == 0)
+	}
+
 	fn insert(&self, tree: usize, key: &[u8], value: &[u8]) -> Result<()> {
 		let tree = self.get_tree(tree)?;
 		let db = self.db.get()?;
@@ -1,7 +1,7 @@
 use crate::*;

 fn test_suite(db: Db) {
-	let tree = db.open_tree("tree").unwrap();
+	let tree = db.open_tree("tree:this_is_a_tree").unwrap();

 	let ka: &[u8] = &b"test"[..];
 	let kb: &[u8] = &b"zwello"[..];
@@ -14,7 +14,7 @@ fn test_suite(db: Db) {

 	assert!(tree.insert(ka, va).is_ok());
 	assert_eq!(tree.get(ka).unwrap().unwrap(), va);
-	assert_eq!(tree.len().unwrap(), 1);
+	assert_eq!(tree.iter().unwrap().count(), 1);

 	// ---- test transaction logic ----

@@ -148,3 +148,15 @@ fn test_sqlite_db() {
 	let db = SqliteDb::new(manager, false).unwrap();
 	test_suite(db);
 }
+
+#[test]
+#[cfg(feature = "fjall")]
+fn test_fjall_db() {
+	use crate::fjall_adapter::{fjall, FjallDb};
+
+	let path = mktemp::Temp::new_dir().unwrap();
+	let config = fjall::Config::new(path).temporary(true);
+	let keyspace = config.open_transactional().unwrap();
+	let db = FjallDb::init(keyspace);
+	test_suite(db);
+}
@@ -1,6 +1,6 @@
 [package]
 name = "garage"
-version = "2.0.0"
+version = "2.1.0"
 authors = ["Alex Auvolat <alex@adnab.me>"]
 edition = "2018"
 license = "AGPL-3.0"
@@ -59,6 +59,7 @@ opentelemetry.workspace = true
 opentelemetry-prometheus = { workspace = true, optional = true }
 opentelemetry-otlp = { workspace = true, optional = true }
 syslog-tracing = { workspace = true, optional = true }
+tracing-journald = { workspace = true, optional = true }

 [dev-dependencies]
 garage_api_common.workspace = true
@@ -79,7 +80,7 @@ static_init.workspace = true
 assert-json-diff.workspace = true
 serde_json.workspace = true
 base64.workspace = true
-crc32fast.workspace = true
+crc-fast.workspace = true

 k2v-client.workspace = true

@@ -92,6 +93,7 @@ k2v = [ "garage_util/k2v", "garage_api_k2v", "garage_api_admin/k2v" ]
 # Database engines
 lmdb = [ "garage_model/lmdb" ]
 sqlite = [ "garage_model/sqlite" ]
+fjall = [ "garage_model/fjall" ]

 # Automatic registration and discovery via Consul API
 consul-discovery = [ "garage_rpc/consul-discovery" ]
@@ -103,6 +105,8 @@ metrics = [ "garage_api_admin/metrics", "opentelemetry-prometheus" ]
 telemetry-otlp = [ "opentelemetry-otlp" ]
 # Logging to syslog
 syslog = [ "syslog-tracing" ]
+# Logging to journald
+journald = [ "tracing-journald" ]

 # NOTE: bundled-libs and system-libs should be treat as mutually exclusive;
 # exactly one of them should be enabled.
@@ -160,8 +160,8 @@ impl Cli {
 			.await?;

 		println!(
-			"Purged {} blocks: deleted {} versions, {} objects, {} multipart uploads",
-			res.blocks_purged, res.versions_deleted, res.objects_deleted, res.uploads_deleted,
+			"Purged {} blocks: deleted {} block refs, {} versions, {} objects, {} multipart uploads",
+			res.blocks_purged, res.block_refs_purged, res.versions_deleted, res.objects_deleted, res.uploads_deleted,
 		);

 		Ok(())
@@ -100,7 +100,7 @@ impl Cli {
 						drain_msg = true;
 						"draining metadata.."
 					}
-					_ => unreachable!(),
+					_ => continue,
 				};

 				failed_nodes.push(format!(
--- a/Show More
+++ b/Show More