gsadmin/rustfs
1
0
Fork 0
mirror of https://github.com/rustfs/rustfs.git synced 2026-07-26 08:18:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(ci): widen Days=5 expiration poll window to 8*lc_interval (#4779)

Browse Source 
Follow-up to #4772. After forcing every-cycle ILM evaluation the Days=1
plateau is reliable, but test_lifecycle_expiration / test_lifecyclev2_expiration
still flaked intermittently on the *second* assertion (`assert 4 == 2`):
the Days=5 (expire3) objects were not expired within their poll window.

Cause: _wait_for_lifecycle_count for expire3 starts its N*lc_interval
deadline only after the Days=1 poll returns (~1 debug-day in). With N=5
and lc_interval == debug_day, the 5*debug_day terms cancel and the slack
past the Days=5 due time is only ~1 debug-day (~9s) -- which a single slow
scanner cycle (observed ~13s spacing under CI load) can exceed, leaving the
count stalled at 4. Bumping the two expire3 windows to 8*lc_interval raises
the slack to ~39s, comfortably above the observed cycle jitter.

Test-only, lane-scoped: does not touch RUSTFS_ILM_DEBUG_DAY_SECS or the
4*lc_interval < 5*debug_day plateau invariant. The expire3 target count (2)
is terminal (nothing expires after it), so a wider window can never
over-expire. Also documents the #4772 RUSTFS_DATA_USAGE_UPDATE_DIR_CYCLES=1
knob in lifecycle_behavior_tests.txt.
This commit is contained in:
Zhengchao An
2026-07-12 22:15:05 +08:00
committed by GitHub
parent 418f505a81
commit cf142e7fdd
2 changed files with 15 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/s3-tests/lifecycle_behavior_tests.txt
+13
View File
@@ -20,9 +20,22 @@
# RUSTFS_ILM_DEBUG_DAY_SECS=10 (1 "day" = 10s, matches s3-tests lc_debug_interval)
# RUSTFS_SCANNER_ENABLED=true
# RUSTFS_SCANNER_CYCLE=2 (scan every 2s, well under a debug-day)
# RUSTFS_DATA_USAGE_UPDATE_DIR_CYCLES=1 (re-descend compacted dirs every cycle so
# ILM is evaluated ~every 2s, not once per 16 cycles;
# without this a Days=1 object due at 10s is not
# expired until the next ~32s boundary — #4772)
# RUSTFS_SCANNER_START_DELAY_SECS=0
# RUSTFS_API_STALE_UPLOADS_CLEANUP_INTERVAL=2s (multipart abort lane)
#
# Polling windows (patches/0002): each assertion polls list_objects until the
# expected count is reached or a N*lc_interval deadline elapses. The Days=5
# (expire3) window uses 8*lc_interval rather than 5: its deadline only starts
# after the Days=1 poll returns (~1 debug-day in), so a 5*lc_interval window left
# only ~1 debug-day of slack past the Days=5 due time — a single slow scanner
# cycle (observed ~13s under CI load) could push the expiry just past it and
# stall the count at 4 (assert 4 == 2). 8*lc_interval restores comfortable slack;
# the target count (2) is terminal so a wider window never over-expires.
#
# Point run.sh at this file with IMPLEMENTED_TESTS_FILE=<this path>.
#
# First batch (ilm-10). Each entry was verified against the pinned upstream source
scripts/s3-tests/patches/0002-lifecycle-behavior-polling.patch
+2 -2
View File
@@ -51,7 +51,7 @@ index bf2f2b1..f3ef394 100644
+ keep2_objects = client.list_objects(Bucket=bucket_name).get('Contents', [])
+ expire3_objects = _wait_for_lifecycle_count(
+ lambda: client.list_objects(Bucket=bucket_name).get('Contents', []),
+ 2, 5*lc_interval)
+ 2, 8*lc_interval)
assert len(init_objects) == 6
assert len(expire1_objects) == 4
@@ -76,7 +76,7 @@ index bf2f2b1..f3ef394 100644
+ keep2_objects = client.list_objects_v2(Bucket=bucket_name).get('Contents', [])
+ expire3_objects = _wait_for_lifecycle_count(
+ lambda: client.list_objects_v2(Bucket=bucket_name).get('Contents', []),
+ 2, 5*lc_interval)
+ 2, 8*lc_interval)
assert len(init_objects) == 6
assert len(expire1_objects) == 4