Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-131596ClosedFixed
mongos segfaults disposing a sharded merge cursor with a detached OpCtx
A sharded aggregation crashes mongos during cursor teardown: {{MergeCursorsStage::doDispose()}} calls {{BlockingResultsMerger::kill()}} with a null OperationContext.
*Trigger*
mongos fills a batch with a count-bounded loop:
for (long long objCount = 0; objCount < batchSize; ++objCount)
{ next = ccc->next(); ... }
The crash needs the results to exactly fill the final batch (count == batchSize) with the cursor then at EOF. The loop stops on the count bound without ever pulling the trail
Fixed Versions
9.1.0-rc09.0.0-rc18.3.16
SERVER-135767ClosedFixed
upload_sbom_via_silkbomb_if_changed fails on v8.0-staging: stale evergreen/write_silkbomb_env.sh path
h2. Overview
{{upload\_sbom\_via\_silkbomb\_if\_changed}} fails on {{v8.0\-staging}} \(build variant {{upload\-sbom\-if\-changed}}\) because the "Write temporary AWS credentials to Silkbomb environment file" step still references a deleted script path.
h2. Root cause
SERVER\-129061's legacy\-per\-branch\-SBOM\-automation removal deleted {{evergreen/write\_silkbomb\_env.sh}} on this branch \(replaced by {{evergreen/functions/security\_reporting\_scripts/write\_aws\_creds\_to\_silkbomb\_env\_fi
Fixed Versions
9.1.0-rc08.0.37
SERVER-127788ClosedFixed
Fix mock OCSP responder hang on SIGINT under Python 3.13
the mock OCSP responder process hangs after receiving SIGINT under python 3.13 causing the evergreen suite to hit the timeout
Fixed Versions
9.0.0-rc10118.3.138.0.37
SERVER-127366ClosedFixed
Suppress $_internalFle* operator counters for FLE2 metrics
Components
Queryable Encryption
Fixed Versions
9.0.0-rc10118.3.88.0.37
SERVER-107248ClosedFixed
Make WiredTiger eviction worker threads configurable at runtime
Many of [the server parameters that configure WiredTiger|https://github.com/mongodb/mongo/blob/cc5174f49e293801bc61daca667a4ed7fb702688/src/mongo/db/storage/wiredtiger/wiredtiger_global_options.idl] (both the main instance and the spill instance) are currently only settable at startup and cannot be changed at runtime.
Fixed Versions
8.3.0-rc08.0.37
SERVER-132037ClosedFixed
Resolve bsoncolumn decoder discrepancy in handling 8609800
Fixed Versions
7.0.0-rc119.1.0-rc09.0.48.0.37
SERVER-133954ClosedFixed
Declare missing lock_manager dependencies in Bazel unit-test targets
The Bazel unit tests message_filter_loader_test and message_compressor_test use LockManager symbols without declaring the //src/mongo/db/concurrency:lock_manager dependency.
This caused unit_test_group1 to fail during linking on rhel8-debug-ubsan (BF-45449), including the message_filter_loader_test_with_debug target.
The dependency was added to:
- src/mongo/db/modules/enterprise/src/mfp/BUILD.bazel
- src/mongo/transport/BUILD.bazel
Related PR: https://github.com/10gen/mongo/pull/60525
Validat
Fixed Versions
8.0.37
SERVER-111150ClosedFixed
Use pcre_util::quoteMeta(...) in change streams code
Change streams code currently uses a function {{regexEscapeNsForChangeStream()}} to escape an input string for usage inside regexes.
{{regexEscapeNsForChangeStream}} does not perform any change-stream specific things, so its calls should be replaced by calls to the [already existing {{pcre_util::quoteMeta()}} function|https://github.com/10gen/mongo/blob/c963454373c9ca3dc0a3ddadf950a1ca6499231c/src/mongo/util/pcre_util.h#L67].
Components
Change streams
Fixed Versions
8.3.0-rc08.0.37
SERVER-118451ClosedFixed
ConcurrentTokenAcquisitionWithQueueing unit test assertions are too restrictive
The ConcurrentTokenAcquisitionWithQueueing involves queueing a few threads waiting for a token, advancing the mocked clock by the time it takes to acquire a token, and verifying that a token is dispensed after each advance.
This doesn't quite match the implementation of the underlying token bucket though, which isn't fully synchronized. Queuing is implemented by "borrowing" tokens that don't exist in the token and returning a "nap time" that the borrower has to sleep in order for the borrow
Fixed Versions
8.3.0-rc08.0.37
SERVER-119621ClosedFixed
Skip checking log for find cmd timeout in configurable_fetcher_timeout.js
Components
Replication
Fixed Versions
8.3.0-rc08.0.37
SERVER-125747ClosedFixed
Improve SessionCatalog interface
Currently the SessionCatalog interface, specifically scanSessions is overly broad on what it allows callers to do in terms of state mutation. This function can be spit out into the following:
killSessions: which kills sessions with a given predicate
findExpiredParentSessions: Returns the set of parent session IDs whose last checkout time is earlier than some value
scanSessionsReadOnly: scans a read only list of sessions and provides the caller a const view of an ObservableSession.
Fixed Versions
9.0.0-rc08.3.138.0.37
SERVER-87919ClosedFixed
Stop taking unstable checkpoints during recovery for restore
After SERVER-85688, we are advancing our stable timestamp as we apply oplog entries during restore. This should mean that we can change from taking unstable checkpoints to taking stable checkpoints. This should also apply for magic restore.
Components
Replication
Fixed Versions
9.0.0-rc08.3.98.0.37
SERVER-113344ClosedFixed
bulkWrite returns success items despite errorsOnly:true on write without shard key
When issuing a *bulkWrite* with *errorsOnly:true* and the operation goes through the two-phase write ([write_ops/write_without_shard_key_util.cpp|https://github.com/10gen/mongo/blob/dceb2114550d5c3c50ae77bac76c8828ecb2908e/src/mongo/s/write_ops/write_without_shard_key_util.cpp#L152]) without shard key path (because the filter doesn’t contain the shard key and the write requires an upsert), the server returns a cursor containing per-item success reply items. This violates the contract of {*}error
Fixed Versions
8.3.0-rc08.0.37
SERVER-133696ClosedFixed
Use $count instead of $out in agg_match.js
The test simply asserts that half the collection's documents have flag: false. No need for a separate collection. This may also unblock running the test in sharded suites.
Fixed Versions
9.1.0-rc08.3.139.0.48.0.37
SERVER-96367ClosedFixed
Increase timeout for change_stream_optimization_fuzzer on enterprise-rhel-8-64-bit-dynamic-classic-engine
Fixed Versions
8.1.0-rc08.0.37
SERVER-117624ClosedFixed
Long running index builds may create temporary unavailability upon rename in sharded clusters
The rename DDL coordinator retries indefinitely whenever errors happen during the commit phase (SERVER-61416) . While this ensures that the rename operation eventually completes, it may pose a temporary unavailability risk.
After acquiring the critical section (blocking CRUDs), if an index build is ongoing on the *source* collection on one or more shards, the operation will [fail|ttps://github.com/mongodb/mongo/blob/eb58aeaa87059dce6d28715c1d9cc7522bc4cff1/src/mongo/db/shard_role/shard_catalo
Fixed Versions
9.1.0-rc09.0.0-rc18.0.37
SERVER-103946ClosedFixed
Tag max_doc_size.js to stop failures in slow variants
This test occasionally fails in certain passthroughs on TSAN variants since some of the operations take so long to complete. We should tag the test to avoid these spurious failures.
Fixed Versions
8.2.0-rc08.0.37
SERVER-100548ClosedFixed
Use dynamic port for auditclient test
In audit_client_observer_test.cpp we use a fixed port. In some cases this port may collide to one in use.
We should let Asio set itself one and then use that one in the testing.
Fixed Versions
8.1.0-rc08.0.37
SERVER-135389ClosedFixed
Replace std::ranges::find with std::find in session_workflow_test.cpp in v7.0
Fixed Versions
7.0.48
SERVER-98105ClosedFixed
Pause drop pending ident reaper until backup cursor is closed
As part of SPM-3810, we identified that issues were occurring due to backups happening at the same time as a drop command was issued. To avoid this issue, we can identify whenever we are taking a backup and pause the drop pending ident reaper until we are done with the backup and have closed the cursor.
Fixed Versions
8.1.0-rc08.0.347.0.48