Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-136534ClosedFixed
Add internalEnablePathArraynessTraverseFElision query execution knob
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-133956ClosedFixed
Support higher shard_count for resmoke_suite_test
Allow setting `shard_count` for a `resmoke_suite_test` beyond the arbitrary limit of 50 typically enforced by Bazel. A config transition of `--test_sharding_strategy=forced=N` is applied when set above 50.
Updates the bazel wrapper so that any `--test_sharding_strategy` passed via CLI disables the config transition and respects the user-provided strategy.
Updates a couple scripts that process the bazel output directory to account for artifacts being found under the different configuration
Components
Testing Infrastructure
Fixed Versions
9.1.0-rc09.1.0-rc10229.0.7
SERVER-136464ClosedFixed
Isolate server and crypt package artifacts
Fixes the artifact collision behind [BF-46681](https://jira.mongodb.org/browse/BF-46681). `package` and `crypt_lib_package` uploaded their `Packages` archives to the same S3 key, allowing crypt-only packages to replace the server RPM/DEB archive consumed by SELinux, package testing, publishing, and release staging.
Move the crypt archive and its publishing consumer under `${project}/mongo_crypt/...`, preserving the existing server archive path. Add `--package-kind` to `package_test.py` and its
Components
Bazel
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-137249Open
[v9.0] Preserve LLD 19.1.7 for PPC64LE cross builds
The v9.0 PPC64LE cross-compilation backport (SERVER-136667, https://github.com/10gen/mongo/pull/64292, commit 6240b023a57b197466df8ad1fc04265cd1d37d75) selects the ARM64-hosted LLD bundled with the published PPC GCC cross archive. That archive revision, 1bce2cbf33a4c08af2e73c3b2084e282ca76f3f5, contains MongoDB LLD 23.1.0, whereas native v9.0 PPC release builds use the branch's pinned v5 LLD 19.1.7.
PPC is intended to use LLD. Preserve its existing linker version when moving compilation to ARM6
Components
Build
Fixed Versions
9.0.7
SERVER-134400ClosedFixed
Add adaptive retry to change_stream_future_start_time_new_shard
Fixes BF-46061: change_stream_future_start_time_new_shard now retries once with an adaptive budget derived from measured setup latency, instead of failing the setup assert when addShard/moveChunk is too slow on a host.
Fixed Versions
9.1.0-rc09.1.0-rc10229.0.7
SERVER-134546ClosedFixed
Fix flaky BloomFilterWithStringKeys unit test
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-134664ClosedFixed
maxEstimatedScanBytes query rejection is not logged at default verbosity
h2. Summary
When a query is rejected because it requires an unbounded COLLSCAN on a collection exceeding the {{maxEstimatedScanBytes}} threshold (SERVER-127688), no log line is emitted at default verbosity to indicate the rejection happened. The only trace in the mongod log is the generic user-assertion log line under the {{ASSERT}} component, which is only emitted at {{LOGV2_DEBUG}} level 1 and thus invisible unless verbosity is explicitly raised (e.g. {{{}db.setLogLevel(-5, "assert"){}}}).
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-126814ClosedFixed
$bucket on untracked viewless timeseries can optimize the pipeline on mongos before prepending $_internalUnpackBucket
For timeseries collections, timeseries rewrites (prepending $_internalUnpackBucket) must happen before any optimizations. On mongos we optimize the pipeline if one of the following are true:
* routingTableIsAvailable
* requiresCollationForParsingUnshardedAggregate
* hasChangeStream
* shouldDoFLERewrite
If routingTableIsAvailable is true, then we can perform the timeseries rewrite on mongos and should optimize the pipeline correctly.
If routingTableIsAvailable is false then we do
Affected Versions
9.0.0-rc0
Fixed Versions
9.1.0-rc09.0.7
SERVER-135622ClosedFixed
Disable lookup unwind hash join
Details in this [doc.|https://docs.google.com/document/d/1IQB_qyAspG6URlilsJ4Q0e2A-uztL2uobL8-IzC_CDU/edit?tab=t.k74ri3a7fjzr#heading=h.d3y1w5bsedh]
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-136923ClosedDone
Use target architecture in cross-compiled server buildInfo
h2. Problem
s390x release binaries built on ARM64 embed the execution host architecture in buildInfo: distarch and target_arch are "aarch64". This causes jstests/sharding/query/pipeline_length_limit.js to expect 1000 stages while the correctly compiled s390x server returns 700.
Confirmed on v8.3-staging at 2a59d93a96806679bb14301d968498441771a294 and v9.0-staging at 78e4e61309ddf75f9596bfcca5bf969e67d812f2; tracked by [BF-46763|https://jira.mongodb.org/browse/BF-46763]. The affected cross-compi
Fixed Versions
8.3.169.0.78.0.37
SERVER-135406ClosedFixed
Disable change stream multiversion tests
Temporarily disable change stream v2 tests inside `sharding_multiversion`, to resolve [BF-46436](https://jira.mongodb.org/browse/BF-46436).
Components
Change streamsSharding
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-136595ClosedFixed
Update PGO profiles
Updated PGO profile numbers for performance.
Fixed Versions
9.0.7
SERVER-132633ClosedFixed
Retry replSetStepUp in remove_newly_added_field_after_finishing_initial_sync.js
h3. Problem
jstests/replsets/remove_newly_added_field_after_finishing_initial_sync.js's local {{stepUpNode()}} helper calls {{replSetStepUp}} once with no retry. A real election can legitimately fail even after its dry run succeeds (a "spoiled vote" when the responder's concurrent stepdown interrupts its {{LastVote}} write — see {{ReplicationCoordinatorImpl::processReplSetRequestVotes}}, already covered by {{NodeFailsVoteRequestIfItFailsToStoreLastVote}}). This is exactly why {{ReplSetTest.st
Fixed Versions
9.1.0-rc09.0.7
SERVER-137085ClosedFixed
Update PGO profiles
Updated PGO profile numbers for performance.
Fixed Versions
9.1.0-rc09.0.7
SERVER-136570ClosedFixed
Add multiversion exclusion for timeseries_empty_metafield_write.js
Fixed Versions
9.0.7
SERVER-134070ClosedFixed
Release held intents when locks are yielded
In SERVER-124365 we extended the lifetime of intents to match their underlying locks in WuoW. There is now a gap where when locks are yielded during long running operations we do not release intents. This can cause issues with state transitions if an operation yields its locks but continues to hold a write intent
Fixed Versions
9.1.0-rc09.1.0-rc10229.0.7
SERVER-135656ClosedFixed
Add metrics for change streams shard-targeting V2 state machine, control events, and placement history
h2. Problem
The V2 shard-targeting state machine (SPM-1941) is a 9-state FSM running on mongos, but we have no visibility into its runtime behavior. We currently expose cursor/throughput/error metrics, but nothing tells us whether the FSM itself is healthy, whether it's stuck in degraded mode, how much topology churn it's processing, or whether placement history lookups (which the FSM depends on) are fast. This matters for backporting to 9.0, where we'll want fast triage signals without needing
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-136302ClosedFixed
Retain traverseF in SBE $elemMatch MatchFrames
_Filed by the AI query bug hunter (qhunt), an agent that reviews query code for corner-case bugs. A differential oracle and the repro below confirmed the bug._
h2. Problem
Inside an object {{$elemMatch}}, SBE checks path arrayness for the child's element-relative path against the root document. If the collection has a non-multikey index on a root field with the same name, SBE skips array traversal for that field inside the array elements. Elements whose field is an array then never match. The q
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7
SERVER-137133ClosedFixed
wait for resharding cleanup and replication before comparing magic restore dbhashes
This is the same fix as [SERVER-132602|https://github.com/10gen/mongo/pull/60026/changes#top] for remaining magic restore jstests.
Fixed Versions
9.1.0-rc09.1.0-rc10259.0.7
SERVER-136507ClosedFixed
Path arrayness constraints are not re-checked on PlanExecutor restore
We recheck path arrayness constraints on yield, but not plan executor restore. This creates a situation where a query relying on path arrayness constraints may read from a snapshot where those constraints are violated.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.7