ci(kms): add a nightly live-Vault lane and stop leaking behavior keys (#5999)

No workflow ever set RUSTFS_KMS_VAULT_TOKEN, so live_vault_backends() returned an empty set in every CI run and behavior_rotation.rs never asserted the working half of rotate/versioning; the #[ignore] live-Vault tests had never executed in CI either. nightly-gnu.yml gains a kms-vault-lane job (vault server -dev with KV2 + Transit, full rustfs-kms suite with the lane on, the dev-Vault ignored tests, and the AppRole live script) plus a separate kms-vault-ha-failover job for the three-node Raft failover script, isolated so an election-timing flake cannot mask the main lane's verdict. GitHub-hosted ubuntu-latest rather than the self-hosted fleet: the HA script needs Docker, and e2e-s3tests.yml's banner records how the heterogeneous sm-standard pods burned the last docker-dependent workflow.

The behavior harness now records every key TestKms::create_key mints and deletes them after each Vault-backed for_each_backend case, on a fresh manager over the same configuration with the immediate-deletion gate enabled for cleanup only. Transit needs the deletion issued twice (first call parks the key in PendingDeletion, the second destroys it); KV2 destroys on the first call. Verified against a real dev Vault: after a full suite run the server holds zero behavior-* keys.

Also fixes test_vault_cancel_key_deletion_persists_state, which was broken by construction — Default::default() never picks up the insecure-dev-defaults env override, so the HTTP dev Vault the test requires was always refused. It now declares development mode on the config, and passes.

Refs rustfs/backlog#1774, rustfs/backlog#1562.
This commit is contained in:
Zhengchao An
2026-08-12 23:14:04 +08:00
committed by GitHub
parent 1021d7228a
commit 16d381fc0e
3 changed files with 229 additions and 4 deletions
+139
View File
@@ -55,3 +55,142 @@ jobs:
- name: Build RustFS
run: cargo build --release --locked --target x86_64-unknown-linux-gnu -p rustfs --bins
# Live-Vault lane for the rustfs-kms suite (rustfs/backlog#1774).
#
# RUSTFS_KMS_VAULT_TOKEN is the single switch that adds the Vault KV2 and
# Vault Transit backends to every for_each_backend spec in
# crates/kms/tests/behavior_*.rs (see crates/kms/AGENTS.md). rotate and
# versioning are advertised only by the Vault backends, so without this lane
# no CI run ever asserts the working half of behavior_rotation.rs — a
# rotation that silently dropped historical key versions would stay green.
# The same lane runs the dev-Vault #[ignore] tests and the two self-hosting
# live scripts (AppRole login, three-node Raft leader failover).
#
# GitHub-hosted ubuntu-latest, deliberately not the self-hosted sm-standard
# fleet: the HA failover script needs a working Docker daemon, and the
# self-hosted fleet is heterogeneous — a docker-dependent workflow has been
# burned by it before (see the banner in e2e-s3tests.yml, rustfs/backlog#1149).
kms-vault-lane:
name: KMS live Vault lane
runs-on: ubuntu-latest
timeout-minutes: 90
env:
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
# Root token of the ephemeral loopback dev server. Not a secret: the
# server lives only for this job, listens on 127.0.0.1, and holds only
# keys the tests create. The literal value matters — the dev-Vault
# #[ignore] fixtures in crates/kms/src/backends/vault.rs hardcode it.
VAULT_LANE_TOKEN: dev-only-token
VAULT_LANE_ADDR: http://127.0.0.1:8200
# Keeps a runner-level proxy from swallowing the loopback dev-server
# traffic (see crates/kms/AGENTS.md). Actions env keys are
# case-insensitive, so only the uppercase form is set; reqwest reads
# either casing.
NO_PROXY: 127.0.0.1,localhost
steps:
- name: Checkout main branch
uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
with:
persist-credentials: false
ref: main
- name: Setup Rust environment
uses: ./.github/actions/setup
with:
# Dedicated key: rust-cache cannot tell runner images apart, so
# sharing a key with an sm-standard lane would let two different
# system images overwrite each other's artifacts (same reasoning as
# ci.yml's ci-uring lane). Saved from this nightly job itself so the
# next night starts warm.
cache-shared-key: kms-vault-lane
cache-save-if: 'true'
install-build-packaging-tools: 'false'
install-test-tools: 'false'
- name: Install Vault CLI
run: |
set -euo pipefail
wget -qO- https://apt.releases.hashicorp.com/gpg | sudo gpg --dearmor -o /usr/share/keyrings/hashicorp-archive-keyring.gpg
echo "deb [signed-by=/usr/share/keyrings/hashicorp-archive-keyring.gpg] https://apt.releases.hashicorp.com $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/hashicorp.list >/dev/null
sudo apt-get update -qq
sudo apt-get install -y -qq vault
vault version
- name: Start Vault dev server with KV2 and Transit engines
run: |
set -euo pipefail
nohup vault server -dev \
-dev-root-token-id="${VAULT_LANE_TOKEN}" \
-dev-listen-address=127.0.0.1:8200 >/tmp/vault-dev.log 2>&1 &
for _ in $(seq 1 60); do
if curl -fsS "${VAULT_LANE_ADDR}/v1/sys/health" >/dev/null 2>&1; then
break
fi
sleep 1
done
curl -fsS "${VAULT_LANE_ADDR}/v1/sys/health"
export VAULT_ADDR="${VAULT_LANE_ADDR}" VAULT_TOKEN="${VAULT_LANE_TOKEN}"
# Dev mode mounts KV v2 at secret/ by default; Transit is explicit.
# Prove both engines actually work rather than assuming the defaults.
vault secrets enable transit
vault kv put secret/rustfs-ci-lane-probe value=ok >/dev/null
vault kv get secret/rustfs-ci-lane-probe >/dev/null
vault write -f transit/keys/rustfs-ci-lane-probe >/dev/null
- name: Run rustfs-kms suite with the Vault lane on
env:
RUSTFS_KMS_VAULT_TOKEN: ${{ env.VAULT_LANE_TOKEN }}
RUSTFS_KMS_VAULT_ADDR: ${{ env.VAULT_LANE_ADDR }}
run: cargo test -p rustfs-kms --locked
- name: Run dev-Vault ignored tests
env:
RUSTFS_KMS_VAULT_TOKEN: ${{ env.VAULT_LANE_TOKEN }}
RUSTFS_KMS_VAULT_ADDR: ${{ env.VAULT_LANE_ADDR }}
# Filters select the dev-Vault-only #[ignore] tests. The AWS #[ignore]
# tests (backends::aws, service_manager) stay excluded — they need real
# AWS credentials and create billable keys. The AppRole and HA #[ignore]
# tests are excluded here because their own scripts below provision the
# Vault topology they need.
run: |
set -euo pipefail
cargo test -p rustfs-kms --locked --lib backends::contract_tests -- --ignored
cargo test -p rustfs-kms --locked --lib backends::vault -- --ignored
cargo test -p rustfs-kms --locked --test vault_fault_injection -- --ignored
- name: Run AppRole live checks (self-hosting ephemeral Vault)
run: bash scripts/test/vault_approle_kms_live.sh
- name: Show Vault dev server log on failure
if: failure()
run: tail -n 200 /tmp/vault-dev.log || true
# Three-node Raft leader failover (crates/kms/tests/vault_ha_failover_live.rs,
# first validated by rustfs/rustfs#5653). Its own job so an election-timing
# flake cannot mask the main lane's verdict, and vice versa. The script
# provisions and tears down its own Docker cluster.
kms-vault-ha-failover:
name: KMS Vault HA failover lane
runs-on: ubuntu-latest
timeout-minutes: 60
env:
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
NO_PROXY: 127.0.0.1,localhost
steps:
- name: Checkout main branch
uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
with:
persist-credentials: false
ref: main
- name: Setup Rust environment
uses: ./.github/actions/setup
with:
cache-shared-key: kms-vault-lane
cache-save-if: 'false'
install-build-packaging-tools: 'false'
install-test-tools: 'false'
- name: Run HA leader failover live checks (three-node Raft cluster in Docker)
run: bash scripts/test/vault_ha_kms_live.sh