[fix] flaky cases

[warrick1016]

Squashed commit range (inclusive):

• 82fe4fc [fix] tests/gtid/xsync.tcl WRONGTYPE
• 116b192 [fix] tests/gtid/xsync.tcl
• 4251b36 fix: swap_data_comp set comparison uses hardcoded key and wrong sismember args
• 96956f4 [fix] add dbsize_loadsafe
• 467477b [fix] wait for async eviction to finish
• cf259bb [fix] swap_error
• d9a04fd [fix] ci-repeat
• ac423b9 [fix] tests/swap/unit/swap_thread.tcl
• f338755 [fix] add branch for ci-repeat
• 389d01e [fix] obuf-limits
• 9112512 [fix] tests/integration/logging.tcl
• 3f588da [fix] tests/unit/io_thread.tcl: wait for scale-up before verifying log
• 0b5cc37 [fix] tests/unit/io_thread.tcl: avoid re-querying io_thread_2 after snapshot
• 7941433 [fix] tests/integration/rdb.tcl: tolerate slow COW reporting under ASAN
• f1c5da8 [fix] tests/swap/unit/dirty.tcl && import_mode.tcl
• 4fd599a [fix] ci.yml
• c9ac92a [fix] *** [err]: repl mode change propagates to all replicas in tests/gtid/xsync.tcl gtid didn't sync in time
• 90ce0b8 Revert "[fix] *** [err]: repl mode change propagates to all replicas in tests/gtid/xsync.tcl"
• a715faa [fix] tests/gtid/xsync.tcl plan A
• 7d5a7a8 [fix] xsync.tcl fallback to fullresync on wrong type
• cf431a0 [fix] tests/gtid/xsync.tcl passive dr
• 1a9e922 [fix] tests/integration/rdb.tcl failed bgsave prevents writes
• 1ed12d1 [fix] tests/gtid/sync.tcl PARTIAL SYNC SET
• 258b7e2 [fix] ci.yml
• 3207ebb [fix] 5 files with loading risk
• 886daa1 Repair PR15 monitor dispatch
• 2c84e35 [fix] tests/gtid/sync.tcl rewrite multi cases with wait_for_gtid_sync
• f3ba298 [fix] tests/swap/ported/integration/replication.tcl Loading DB in memory
• d86b746 [fix] tests/integration/replication-rdbchannel.tcl Background transfer error
• f195738 [fix] [err]: diskless no replicas drop during rdb pipe in tests/integration/replication.tcl rdb child didn't terminate
• 33ee17e [fix] tests/integration/redis-cli.tcl modification for client time out
• 47fb5bb fix: use __lsan_ignore_object in objectMetaSetPtr to suppress false leak
• f16a290 [fix] *** [err]: swap_thread in tests/swap/unit/swap_thread.tcl Expected '0' to be equal to '1' (context: type eval line 52 cmd {assert_equal [$master swap.debug thread auto-scale-down] 1} proc ::test)
• de33219 Switch PR15 to passive polling
• 95bdd53 [fix] *** [err]: PSYNC2: Partial resync after Master restart using RDB aux fields with expire in tests/gtid/psync2-master-restart.tcl Expected [status ::redis::redisHandle10 master_repl_offset] == [expr 413279+96835] (context: type eval line 49 cmd {assert {[status $master master_repl_offset] == [expr $offset+$expired_offset]}} proc ::test)
• 45caf70 Restore PR15 comment permission
• 162562f [fix] [err]: Test replication partial resync: ok after delay (diskless: no, disabled, save: {1 1}, reconnect: 1) Expected '679' to be equal to '680' (context: type source line 497 file /home/runner/work/Redis-On-Rocks/Redis-On-Rocks/tests/swap/support/util.tcl cmd {assert_equal [$r1 dbsize] [$r2 dbsize]} proc ::swap_data_comp level 2)
• f9270b0 [fix] *** [err]: Test replication with parallel clients writing in different DBs in tests/swap/ported/integration/replication-4.tcl Master - Replica inconsistency, Run diff -u against /tmp/repldump*.txt for more info
• 61cffd0 [fix] skip backlog overflow assert in SWAP mode for psync2-master-restart
• 5428e7e [fix] replace hard after-delay with wait_for_condition in gtid expire test
• 6e90348 [fix] replace hard after-delays with deterministic waits in dirty.tcl
• dc61dcc [fix] replace hard after-delays with deterministic waits in expire.tcl
• 5936705 [fix] replace hard after-delays with deterministic waits in absent_cache.tcl
• 9754bbe fix: stabilize ttl_compact flushdb test
• 464f3af test: fix flaky SCAN with expired keys in swap mode
• 7522fa3 test: increase replication initial sync timeout in gtid/replication-psync
• 7a47118 [fix] avoid time-out of swap or swap-asan
• 0b3e548 [fix] revert zset evict-only-dirty-subkeys to single-evict pattern
• deac80e [fix] increase wait_for_log_messages timeout for ASAN in diskless replication test
• a07415d [fix] replace hard after-delays with deterministic waits in swap ported integration tests
• 8e2afec test: suppress SWAP SST-AGE-LIMIT pings in replication stream captures
• 598fa51 test: reduce replication.tcl test time in swap mode
• 1e00201 Restore PR15 active CI polling
• e911e22 test: fix flaky SIGALRM stack trace test in swap mode
• c91e5c4 [fix] replication-rdbchannel.tcl: increase timeout for rdb-channel abort test
• d357ad6 [fix] integration tests: replace hard waits with wait_for_condition
• f5e6651 Backfill recent PR15 CI completions
• 6fc5e30 test: increase repl-timeout for SWAP+diskless sync in replication-psync
• fc8f0dd [fix] checkpointDirPipeWriteHandler
• af298b9 [fix] *** [err]: Test replication partial resync: backlog expired (diskless: yes, disabled, reconnect: 1) in tests/gtid/replication-psync.tcl Slave not correctly synchronized
• c1aa42c Bootstrap latest PR15 head CI
• f8a75e6 test: skip debugCommand backtrace check in ASAN mode
• f2412e2 fix: increase ASAN timeout for diskless-load swapdb replication test
• e44b60c [fix] backlog expired (diskless: yes, swapdb) in swap/ported/integration/replication-psync.tcl
• 5677fed [fix] replication-rdbchannel: verify replica disconnects before waiting for error log
• bad8dbb fix: increase ASAN timeout for diskless rdb pipe bgsave child wait
• 08dc0c0 [fix] rio.c: detect all-failed connections in last rioConnsetWrite iteration
• fba9236 [fix] replication-rdbchannel: restore 50s bgsave-start timeout after polling interval change
• c2d3e75 Fix PR15 workflow-name matching
• 1202718 [fix] replication-psync: reduce write load under ASAN to fix diskless full-sync livelock
• 67e0171 [fix] replication-psync: raise ASAN bg_limit from 2000 to 5000 ops
• 2aff050 fix: increase ASAN read_cli timeout for i-search tests
• 381beef Limit PR15 comments to mem and swap
• 41988fe Fix intermittent 'Generate stacktrace on assertion with user data hidden' failure in logging.tcl
• 42a0994 [fix] swap_error: scope rio-error injection to specific RIO actions to fix intermittent failures
• 55d3f2c fix: increase read_cli timeout for db_num reconnect test
• 6c458d5 [fix] tests/gtid/xsync.tcl: replace after 100 with wait_load_handlers_disconnected
• eeb1d15 fix: skip unnecessary RocksDB pre-flush in swap_replicationStartPendingFork when swap-repl-rordb-sync is disabled
• 88b4183 [fix] swap_thread in tests/swap/unit/swap_thread.tcl: use MULTI/EXEC to prevent background timer race
• d931203 fix: 'Test master aborts rdb delivery if all replicas are dropped'
• 02f772f [fix] Fix flaky test: Clients are evenly distributed among io threads
• 9d6ed31 [hotfix] fix flaky test: not enough good replicas state change during long script
• 159a429 Fix flaky test: Temp rdb will be deleted if we use bg_unlink when shutdown
• d8bd014 fix: reduce numclients from 16 to 8 to prevent memory pressure on 7GB CI runners
• bc8d6e0 Dispatch fresh PR15 runs instead of reruns
• c6130d4 fix: increase gtid-sync retry for swap mode in 'repl mode change propagates' test
• 39884c3 fix: increase read_cli idle timeout for i-search tests in swap mode
• 02a9172 [fix] corrupt-dump-fuzzer.tcl: prevent OOM crash from oversized SETRANGE during fuzzy traffic
• 015ef10 fix: tests/integration/replication-rdbchannel.tcl Test master aborts rdb delivery if all replicas are dropped Background transfer error
• 1198f17 fix: tests/integration/replication-rdbchannel.tcl Test master aborts rdb delivery if all replicas are dropped shutdown nosave
• 5cf6e64 [hotfix] fix flaky test: psync2-pingoff LOADING race on replica
• cf90a59 Filter stale jobs from repeated CI attempts
• 3e446ca [hotfix] fix flaky test: restart master (yes => yes) slave (no) LOADING race
• 08b06e6 fix(test): sync diskless all replicas drop during rdb pipe in replication.tcl
• 6b29c8f Filter PR15 attempt jobs by run start time
• daac6c5 fix(test): stabilize tests/unit/scan.tcl expired SCAN cases
• 3ed625e fix(test): stabilize tests/swap/unit/slowlog.tcl slave slow log
• 5eff835 fix: tests/integration/replication.tcl diskless no replicas drop during rdb pipe rdb child didn't terminate
• af1e8c3 fix(test): stabilize swap replication startup waits
• 9ff91cc fix: tests/integration/replication-rdbchannel.tcl Test master aborts rdb delivery if all replicas are dropped Background transfer error
• 87533a4 fix: tests/integration/replication-rdbchannel.tcl Test master aborts rdb delivery if all replicas are dropped Background transfer error
• a4cb348 Remove CI Repeat workflow
• fc1af3f Backstop proactive PR15 dispatch recording
• f7d3c10 test: stabilize tests/gtid/master_restart.tcl and tests/gtid/replication-psync.tcl
• 78b72fb test: stabilize tests/gtid/sync.tcl dbsize waits
• 7732212 test: restore start_write_load compatibility for gtid/xsync.tcl
• ef9d660 fix(test): extend swap SCAN evicts wait in tests/unit/scan.tcl
• 4b17fbd fix(test): improve tests/gtid/xsync.tcl gtid sync timeout diagnostics
• 954d696 fix(test): make tests/gtid/xsync.tcl repl mode change propagates to all replicas self-contained
• 3f973ca fix: tests/integration/replication-rdbchannel.tcl Test master aborts rdb delivery if all replicas are dropped Background transfer error timing
• 417964e Poll PR15 monitor every 15 minutes
• 97d9f51 fix: tests/integration/replication-rdbchannel.tcl Test master aborts rdb delivery if all replicas are dropped log message of 'Background transfer error' not found diagnostics
• f3a408d try more times for diskless replicas drop during rdb pipe
• c5cc18b .github/workflows/ci.yml: replace fixed -j8 with mem/swap adaptive jobs
• 2e3c083 Merge branch 'fix/loading' into fix/unstable
• 6ac2c44 [fix] debug log fix
• dac172b tests/swap/integration/replication.tcl: stabilize 'shift replid will be defered untill previous master client drain' after timeout waiting replid shift defer done
• 6698062 Revert "tests/swap/integration/replication.tcl: stabilize 'shift replid will be defered untill previous master client drain' after timeout waiting replid shift defer done"
• 8f56efa fix: stabilize swap integration failures

squash: merge commits 7cbafdb..753da44

Squashed commit range (inclusive):

• 7cbafdb fix(src/config.c): support CONFIG SET swap-batch-limit in swap scale-down test
• 22aaedf fix(src/config.c): validate swap-batch-limit atomically
• 3ec39d0 fix: stabilize swap integration flaky cases
• f5abbf0 tests/integration/replication-rdbchannel.tcl: diagnose missing 'Background transfer error' in aborts-rdb-delivery
• f881e23 tests/integration/replication-rdbchannel.tcl: split assertions for aborts-rdb-delivery flaky case
• b5e72d4 fix tests/test_helper.tcl tests/integration/redis-cli.tcl tests/support/server.tcl: avoid swap-asan TIMEOUT in GETDEL/GETEX replication stream and redis-cli replica/output waits
• ef357c3 tests/integration/replication.tcl: raise fast-branch timeout for diskless rdb pipe
• d19c22e tests/swap/integration/bgsave.tcl tests/swap/integration/expire_evict.tcl tests/swap/integration/replication.tcl: stabilize flaky cases in debug reload after bgsave swap out string and slave reconnect master will defer without discarding replicated commands
• 07efee6 test: stabilize tests/swap/integration/swap_load.tcl fix all type meta.len mismatch
• e461758 tests/swap/support/util.tcl: fix fix all type meta.len mismatch expected boolean value but got ""
• 17da5bc tests/gtid/xsync.tcl: relax sync_full assertion in repl mode change case
• 6284b50 tests/swap/integration: harden async wait flakes
• 01a5aa6 tests/swap/unit/client.tcl: bound server disconnect wait
• 7e75975 swap/rdb: add dismissObject for hot keys in ENABLE_SWAP path
• d1f89aa swap/rdb: remove dead aofReadDiffFromParent code from swapRdbSaveProgress
• 7feb1cf swap/rdb: revert dismissObject in ENABLE_SWAP path (caused coredump)
• 9d752e5 swap/rdb: add safe dismissObject for pure-hot keys in ENABLE_SWAP path
• 753da44 remove TODO