gsadmin/rustfs
1
0
Fork 0
mirror of https://github.com/rustfs/rustfs.git synced 2026-07-27 08:38:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

test(ecstore): serialize load-sensitive flaky tests via nextest test-group (#4558)

Browse Source 
Two ecstore test groups pass in isolation but flake under the full parallel
nextest suite, producing CI noise on every in-flight PR (backlog #937):

- store::bucket::tests::bucket_delete_* race make_bucket into
  InsufficientWriteQuorum because they share global disk-registry/lock-client
  state with other concurrently running tests.
- bucket_lifecycle_ops::tests::concurrent_resend_same_part_commits_one_generation
  exceeds its already-maxed 60s lock-acquire deadline only when the suite
  saturates disk I/O.

serial_test's #[serial] does not serialize these across nextest's per-test
process boundary. Add a nextest test-group (max-threads = 1) that matches both
groups so they run serialized, removing the load-induced flake. No test or
production code changes; long-term global-singleton isolation is tracked
separately.

Co-authored-by: heihutu <heihutu@gmail.com>
This commit is contained in:
houseme
2026-07-09 03:12:06 +08:00
committed by GitHub
parent 1bd0f49c7a
commit 7c701d9f2c
1 changed files with 25 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.config/nextest.toml
+25
View File
@@ -0,0 +1,25 @@
# nextest configuration for RustFS.
#
# Serialize two known load-sensitive / global-state-sharing ecstore test groups
# so the full parallel nextest suite stops producing spurious failures
# (backlog #937). These tests pass in isolation but flake under the loaded
# parallel run for two distinct reasons:
#
# * store::bucket::tests::bucket_delete_* share process/global state (disk
# registry, lock client) and race make_bucket into InsufficientWriteQuorum
# when run concurrently with other ecstore tests.
# * bucket_lifecycle_ops::tests::concurrent_resend_same_part_commits_one_generation
# asserts a lock-acquire correctness property whose serialized cross-disk
# commits exceed the (already max'd, 60s) acquire deadline only when the
# suite saturates disk I/O.
#
# serial_test's #[serial] attribute does NOT serialize these across runs:
# nextest executes each test in its own process, where the in-process
# serial_test mutex has no effect. A nextest test-group with max-threads = 1 is
# the mechanism that actually serializes across nextest's process boundary.
[test-groups]
ecstore-serial-flaky = { max-threads = 1 }
[[profile.default.overrides]]
filter = 'package(rustfs-ecstore) & (test(concurrent_resend_same_part_commits_one_generation) | test(/^store::bucket::tests::bucket_delete_(mark_delete_marks|purge_removes|default_s3_delete)/))'
test-group = 'ecstore-serial-flaky'