docs: state that MinIO-encrypted objects are not readable by RustFS

Operators evaluating a MinIO migration had no warning that objects MinIO
wrote with SSE-S3, SSE-KMS, or SSE-C cannot be read back. The container
formats interoperate, so the limitation is easy to discover only after
the data has moved.

Document the limitation where a migration decision is actually made:

- minio-file-format-compat.md gains Part C, covering which object classes
  transfer, the three seams that block each SSE mode with file:line
  evidence, the reverse direction, and the current workarounds. It also
  records that the `rio-v2` MinIO sealed-key parser does not close the
  gap: the feature is absent from released artifacts, and the managed-SSE
  detection gate is not feature-gated and returns before the parser runs.
- kms-backend-security.md gains an operator-facing warning next to the
  backend comparison table, since configuring the static backend with
  MinIO's key material looks like it should work and does not.
- s3-compatibility-matrix.md scopes its SSE row to RustFS's own
  round-trip.

The read path treats an undetected MinIO-encrypted object as unencrypted
rather than failing, so all three notes tell operators to verify migrated
objects by content instead of by status code.

Refs rustfs/backlog#1638.
This commit is contained in:
overtrue
2026-08-02 10:43:33 +08:00
parent 8e3e552576
commit 57bcfc9da3
3 changed files with 102 additions and 3 deletions
+77 -2
View File
@@ -26,9 +26,11 @@ Refs rustfs/backlog#580.
- **Migration**: RustFS already ships a one-way importer that reads a legacy
meta bucket and rewrites bucket-metadata + IAM config into the RustFS meta
bucket (`crates/ecstore/src/bucket/migration.rs`).
- **Server-side encryption**: not covered by the above. Objects MinIO wrote with SSE-S3, SSE-KMS, or SSE-C are **not readable by RustFS** in any shipped build. See [Part C](#part-c--server-side-encryption-sse) before planning a migration that includes encrypted objects.
The remaining work is verification breadth and closing per-config parsing gaps,
not a format rewrite.
For unencrypted objects the remaining work is verification breadth and closing
per-config parsing gaps, not a format rewrite. Encrypted objects are a separate,
unsolved axis (rustfs/backlog#1638).
---
@@ -220,6 +222,79 @@ missing piece is a source adapter that points the importer at a MinIO
---
## Part C — Server-Side Encryption (SSE)
Container-format parity does **not** extend to encrypted object payloads. RustFS currently does not support reading objects that MinIO wrote with server-side encryption — SSE-S3, SSE-KMS, or SSE-C. This is true of every released binary and container image. Tracked in rustfs/backlog#1638.
Note the asymmetry with Parts A and B: the `xl.meta` around a MinIO SSE object parses fine, so such objects list, HEAD, and report plausible sizes. Only the payload is unreadable.
### What can and cannot be migrated
| Object class | Readable after moving the drives / copying via S3 | Notes |
|---|:--:|---|
| Unencrypted objects | ✅ | Parts A and B apply. |
| Bucket metadata, IAM config | ✅ | Via the importer, once a `.minio.sys` source adapter exists (see Part B). |
| Bucket-level default-encryption *configuration* | ✅ | The `encryption` config blob round-trips as a blob; it does not make existing ciphertext readable. |
| MinIO-written SSE-S3 objects | ❌ | Seams 1 and 2 below. |
| MinIO-written SSE-KMS objects | ❌ | Seams 1 and 2 below. |
| MinIO-written SSE-C objects | ❌ | Seam 3 below. |
| RustFS-written SSE objects read back by MinIO | ❌ | See "Reverse direction". |
### Where the read path stops
The primitives match — RustFS implements the same DARE V2 stream format and the same object-key derivation and sealing, and a MinIO sealed-key parser exists (`parse_minio_managed_sealed_key`, `rustfs/src/storage/sse.rs:3195`). Three seams above the cryptography still reject MinIO-written objects.
| # | Seam | Evidence |
|---|---|---|
| 1 | The managed-SSE (SSE-S3 / SSE-KMS) read path returns "not encrypted" unless the object's *persisted* metadata carries the S3 response key `x-amz-server-side-encryption`. RustFS writes that key into metadata on PUT; MinIO's internal sealed-key headers alone do not satisfy the gate. | Gate: `rustfs/src/storage/sse.rs:2432`. RustFS write side: `rustfs/src/storage/sse.rs:391-400`. |
| 2 | MinIO's wrapped-DEK blob (`{"aead": ...}`) is neither produced nor accepted. `is_data_key_envelope` classifies that shape as not a RustFS envelope, and `LocalSseDekEnvelope` is `deny_unknown_fields`. | `crates/kms/src/encryption/dek.rs:425`, `:443`; `rustfs/src/storage/sse.rs:2784-2790`. Already documented for the static backend at `crates/kms/src/config.rs:304-308`. |
| 3 | SSE-C detection keys on `x-amz-server-side-encryption-customer-algorithm`, and `contains_managed_encryption_metadata` omits MinIO's SSE-C sealed-key header, so a MinIO SSE-C object matches neither detection branch. The unsealing code it would need is already written. | Detection: `rustfs/src/storage/sse.rs:2059` and `:3178-3184`; the omitted constant is `rustfs/src/storage/sse.rs:124`. Unsealing: `rustfs/src/storage/sse.rs:2236-2245`. |
### How it fails
Seams 1 and 3 both `return Ok(None)`, which the resolver maps to "this object has no encryption material" (`rustfs/src/storage/sse.rs:1356`). The read is therefore treated as a plaintext read and the stored ciphertext is served, rather than the request failing. **Do not rely on a GET error to tell you an object did not migrate** — validate a sample of encrypted objects by content (checksum against the source) before decommissioning the MinIO side.
Seam 2 does surface an error, but only for objects that got past seam 1.
### The `rio-v2` feature does not change this
`rustfs/src/storage/sse.rs` contains MinIO-interop code behind `#[cfg(feature = "rio-v2")]`, which can give the impression that enabling the feature closes the gap. It does not, for two independent reasons.
- The feature is not compiled into anything that ships. `rio-v2` is absent from both `default` and `full` in `rustfs/Cargo.toml:39`, `:48`, `:51`; release binaries are built with no `--features` flag, and the published images install that binary rather than compiling their own.
- Seam 1 is not feature-gated and runs *before* the MinIO parser is consulted (`rustfs/src/storage/sse.rs:2432` precedes `:2458`). Even with `rio-v2` enabled, a MinIO-written managed-SSE object returns at the gate and never reaches `parse_minio_managed_sealed_key`.
The interop harness reflects this. The reader tests are `#[ignore]` (`rustfs/src/storage/minio_generated_read_test.rs:244`, `:250`), the workflow that would run them is disabled at the GitHub Actions level and states in its own header that end-to-end MinIO-to-RustFS SSE interop is not implemented (`.github/workflows/minio-interop.yml:24-29`, `:34-39`), and the fixture suite's scope note says the tests "do not yet validate full plaintext reconstruction from MinIO-written encrypted data" (`crates/rio-v2/tests/README.md:55`).
### Reverse direction
Migrating back is also unsupported. Under `rio-v2` RustFS writes its own DEK envelope into MinIO's sealed-key metadata slots and labels it with MinIO's seal algorithm (`rustfs/src/storage/sse.rs:1830-1852`), so the metadata is MinIO-shaped while the key bytes are not MinIO-openable. Default builds do not populate those slots at all (`rustfs/src/storage/sse.rs:1796-1798`). Treat RustFS-written SSE objects as readable only by RustFS.
### Working around the limitation
Until rustfs/backlog#1638 lands, the options are:
- Decrypt on the MinIO side first: rewrite the affected objects as plaintext (or copy them out through MinIO's S3 endpoint, which decrypts on read) and migrate the plaintext, applying RustFS-side encryption afterwards.
- Copy through the S3 API rather than moving drives: a client that reads from MinIO and writes to RustFS gets plaintext from the source and lets RustFS encrypt with its own KMS. This re-encrypts rather than preserving ciphertext, and costs a full data transfer.
- Leave encrypted objects on MinIO and migrate only unencrypted data.
Inventory the source first — bucket default-encryption settings mean objects can be encrypted without the uploader having asked for it, so "we never set SSE headers" is not sufficient evidence that a bucket has no encrypted objects.
### SSE interop verdict
| Item | Done | Partial | Todo |
|---|:--:|:--:|:--:|
| DARE V2 stream format parity | ✅ | | |
| Object-key derivation / sealing parity | ✅ | | |
| MinIO sealed-key parser exists (behind `rio-v2`) | | ⚠️ | |
| Managed-SSE detection accepts MinIO-written metadata | | | ❌ |
| MinIO `{"aead": ...}` wrapped-DEK parser | | | ❌ |
| SSE-C detection accepts MinIO-written metadata | | | ❌ |
| Read MinIO-written SSE-S3 / SSE-KMS / SSE-C objects end to end | | | ❌ |
| RustFS-written SSE objects readable by MinIO | | | ❌ |
| CI proof of SSE read parity | | | ❌ |
---
## Phased Plan
The format is already close; the plan is verification, a source adapter, and
@@ -50,6 +50,8 @@ The implemented test list currently covers the common object-storage surface:
| SSE-C and selected SSE-KMS edge cases | Supported | `implemented_tests.txt` |
| Selected versioning, object-lock, checksum, CORS, raw request, and conditional write behavior | Supported | `implemented_tests.txt` |
"Supported" for the SSE row means RustFS encrypts and decrypts its own objects. It does not mean RustFS can read objects another implementation encrypted: objects MinIO wrote with SSE-S3, SSE-KMS, or SSE-C are not readable by RustFS today, which matters when migrating. See [MinIO file-format interoperability, Part C](minio-file-format-compat.md#part-c--server-side-encryption-sse) and rustfs/backlog#1638.
## Planned Standard Coverage
These are standard S3 areas that remain planned work and must not be described