Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-130667ClosedFixed
Gate express update cursor reuse underneath a flag
Fixed Versions
9.0.0-rc08.3.18
SERVER-127943ClosedFixed
Resuming an index build does not handle multikey metadata keys in bulk builder for wildcard index
Resuming an index build does not set {{BulkBuilderImpl::_hasMultiKeyMetadataKeys}} when resuming, meaning that for a wildcard index it may incorrectly try to insert a multikey metadata key twice.
Fixed Versions
9.0.0-rc08.3.18
SERVER-137567ClosedFixed
Disable Atlas Stream Processing builds and Evergreen coverage on v8.3
BF-46779 tracks repeated unit-test link failures on atlas-amazon2023-arm64 and enterprise-amazon2023-arm64 in mongodb-mongo-v8.3-staging. Streams unit-test binaries link cached s2n objects that reference EVP_MD_CTX_get_size_ex, while the AL2023.3 link hosts provide OpenSSL 3.0.8. The cached objects were built with newer OpenSSL headers; the exact producer environment has not been verified.
Disable Streams in Bazel for development and release builds. Make the disable_streams and streams_release_
Fixed Versions
8.3.18
SERVER-138063ClosedFixed
Update PGO profiles
Updated PGO profile numbers for performance.
Fixed Versions
9.1.0-rc08.3.18
SERVER-138423ClosedDone
Set compiling_for_test to true for compile_dbtest
To fix https://jira.mongodb.org/browse/BF-47057
Components
Bazel
Fixed Versions
9.0.98.3.18
SERVER-112903ClosedFixed
Capture information about numKeys and numBytes built during an index build
Once this and SERVER-110442 are in, it should be possible to calculate the throughput of index builds (bytes/time)
Fixed Versions
9.1.0-rc09.0.9
SERVER-132375ClosedFixed
Lower writeConflict probability in list_collections_and_indexes_write_conflicts.js
With the current failure probability, some tests can execute for a very long time, especially debug builds run on slower variants. The test still will server its purpose with a much lower probability, and this will prevent us from needing so many hits of the WriteConflictRetry in order for the test to pass.
Fixed Versions
9.1.0-rc09.0.9
SERVER-131540ClosedFixed
Add write conflict retry to listIndexes and listCollections
Both listIndexes and listCollections do shard role acquisitions (autoGet in listCollections) which may result in calls to wiredTiger. In suites which enable WiredTiger failpoints that cause reads to throw write conflict exceptions, this can cause these operations to bubble up WriteConlicts to the user.
In practice, this should not actually be a production issue because read only transactions do not throw WriteConflictExceptions. However, the general guidance is to act as though any access to
Fixed Versions
9.1.0-rc09.0.9
SERVER-132655ClosedFixed
Retry $out which fails due to concurrent FCV upgrade
The check for DBDirectClient tassert [here|https://github.com/10gen/mongo/blob/c0d96f900c78c285df6bd08b07ee8ae484e41a3a/src/mongo/db/service_entry_point_shard_role.cpp#L2099-L2101] implicitly relied on the fact that [errorMayBeRetried|https://github.com/10gen/mongo/blob/6b0667f8b0d765ee1705ddf8249d15a825085833/src/mongo/db/service_entry_point_shard_role.cpp#L2050] must be false if we are in a dbDirectClient. This is because both [handleStaleError|https://github.com/10gen/mongo/blob/6b0667f8b0d76
Fixed Versions
9.1.0-rc09.0.9
SERVER-133797ClosedFixed
Fix index build metric object lifetime bug
Fixes an issue relating to the lifetime of variables used to track the number of bytes inserted/deleted from indexes as part of an index build's drain writes phase. This issue was recently caught with an ASAN failure.
Components
Build
Fixed Versions
9.1.0-rc09.0.9
SERVER-113796ClosedFixed
Capture per-phase index build duration
We would like visibility into the duration of each index build phase. To do this,
# Add metrics to {{indexBuildsSSS}} for each duration.
# Calculate elapsed time for each phase and store the value.
It's convenient that we already have counters for each phase, and each counter is incremented in a function call where we have access to {{{}ReplIndexBuildState{}}}, so we can just calculate the elapsed time at each location where the respective counter is incremented using {{ReplIndexBuildSta
Fixed Versions
9.1.0-rc09.0.9
SERVER-134959ClosedFixed
Disable minidumps on Windows in config_fault.js
This is to fix the Windows failures on master in BF-35376 that are the result of loader lock deadlock. This is similar to the fixes done in SERVER-124431.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-131820ClosedFixed
Shard-side merge of a sharded $search query loses SEARCH_META metadata cursors
h2. Summary
When mongos delegates the merging half of a sharded {{$search}} query that references {{$$SEARCH\_META}} to a shard, the metadata {{$mergeCursors}} that mongos injected into the {{$setVariableFromSubPipeline}} sub\-pipeline does not survive serialization to the merging shard. The shard reparses the sub\-pipeline as the bare {{planShardedSearch}} metaPipeline \(e.g. {{\[$group, $facet, $replaceRoot\]}}\), whose front stage requires an input source, and the first {{doGetNext\(\)}} hits
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-135668ClosedFixed
Pass reshardingUUID in magic_restore_pit_resharding.js and magic_restore_non_pit_resharding.js
After magic restore aborts an in-progress resharding, the rebuilt ReshardCollectionCoordinator on the db-primary shard re-sends _configsvrReshardCollection and starts a fresh resharding (SERVER-131275). The new op creates resharding collections that fail the post-restore dbhash check (BF-44897).
Apply the same workaround as SERVER-131413: pass a user-supplied reshardingUUID to reshardCollection in magic_restore_pit_resharding.js and magic_restore_non_pit_resharding.js, with a TODO(SERVER-1312
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-135731ClosedFixed
Tag clean_shutdown_log.js as multiversion_incompatible
jstests/replsets/clean_shutdown_log.js fails in replica_sets_last_lts (BF-46566) with exit code 62 (needDowngrade). It started running in multiversion suites after the SERVER-133682 v9.0 backport landed and removed it from the backports_required_for_multiversion_tests.yml exclusion list.
The "records nothing after an unclean shutdown" case SIGKILLs a single-node set shortly after initiate. Under useRandomBinVersionsWithinReplicaSet the restart can pick the last-lts binary. The last checkpoint c
Components
Replication
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-135563ClosedFixed
Send CMC on secondaries with majority read concern
CMC on secondaries runs with one of two modes: checkAtPrimaryTimestamp or checkAtSecondaryTimestamp. With checkAtPrimaryTimestamp, the command is sent [with local read concern and afterClusterTime|https://github.com/10gen/mongo/blob/f80f4a97226a692734655da16ffa23a0d0ac09cb/src/mongo/db/global_catalog/metadata_consistency_validation/shardsvr_check_metadata_consistency_participant_command.cpp#L163-L166] to make sure that replication on the secondary has reached the current primary's latest timesta
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-135154ClosedFixed
Enforce the PinnedConnectionTaskExecutor lifetime contract during mongot shutdown
h3. Context
Make mongod shutdown honor the PinnedConnectionTaskExecutor (PCTE) lifetime contract, so a $search cursor created during shutdown cannot outlive its underlying mongot task executor. This closes the invariant failure tracked by AF-21589 (same class as AF-4442 / SERVER-108562).
h3. Scope
# Gate new mongot cursor creation: once search-executor shutdown begins, cursor acquisition fails fast with ShutdownInProgress (convention already documented in search_task_executors.h) so no new PC
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-134726ClosedFixed
Stop persisting auto-compaction state and resuming auto-compaction for RSWB
As part of this ticket, we will no more resume auto-compaction when {{allowDeletions}} transition from {{false}} to {{{}true{}}}. Auto-compaction will be only paused when {{allowDeletions}} transition from {{true}} to {{false. }}Therefore, we can stop persisting the auto-compaction state when auto-compaction is paused because RSWB is enabled.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-134940ClosedFixed
Threat the RSWB command counter as a monotonic value
Currently, when we are rolling back a RSWB state change, we are [decrementing|https://github.com/10gen/mongo/blob/04d26232f8c23d24cf8ef15371b8a5ff8bdca553/src/mongo/db/topology/user_write_block/writes_recoverable_critical_section_service.cpp#L546-L551] the corresponding serverStatus metric. As part of this ticket, we should consider this metric monotonic and therefore skip the metric decrement in case of a rollback.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-136048ClosedDone
Speed up repo cloning for performance tests
As demonstrated in [DEVPROD-42167], the now-available module-specific repo clone depth allows Evergreen tasks to set specific clone depths on their modules. This has demonstrated a significant improvement for performance tests (see [comment|https://jira.mongodb.org/browse/DEVPROD-42167?focusedCommentId=8961018&focusedId=8961018&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-8961018]). Enable this on performance tests to speed them up.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9