Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-136167ClosedFixed
Reduce unnecessary _validateEoo calls as much as possible
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-138467ClosedFixed
Always describe oplog mayTruncateUpTo as an exclusive upper bound
Update code comments to make sure that oplog mayTruncateUpTo is always described as an exclusive upper bound.
The oplog truncation contract is that at least one record with recordId <= mayTruncateUpTo must remain in the oplog after truncation.
Also add a short-circuit to skip checking the persisted timestamp when all records are pinned.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-135983ClosedFixed
Implement inference for ExpressionConstant and ExpressionFieldPath
Implement getOutputType for these two. It should be trivial after SERVER-135978.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-131927ClosedFixed
Capture cursor read failures on mongod in query stats
Sub-ticket for SERVER-87657.
Handle getMore failures for registered find/aggregate/change-stream cursors on mongod, where the query-stats key has already moved from OpDebug onto ClientCursor. Thread an errored flag through cursor dispose, set that flag in the getMore catch/error path, and pass it into writeQueryStatsOnCursorDisposeOrKill so the error is recorded during cursor teardown.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137550ClosedDone
Detect when a cluster needs to recursively merge during an index build
When spilling during an index build, if the number of range iterators exceeds a certain limit we perform recursive merging on the sorted ranges, which is expensive and leads to longer lasting index builds. If we can detect when we are performing recursive merges, we will have signal that the cluster is in a sense "overloaded" and that by upgrading the instance size of the cluster we can avoid recursive merging and decrease the time spent on the index build
SERVER-137560ClosedFixed
[8.3] Add boolean_simplifier_tautology.js and bson_size_limit_agg.js to backport test exclusion list
Specifically to the last-patch.all section
Fixed Versions
8.3.18
SERVER-139195Needs Scheduling
Remove feature flag for samplingCE keystring estimation
SERVER-134637 introduced a feature flag for samplingCE keystring estimation, used for hashed/wildcard/collated indices.
We should eventually remove it once it is enabled in the last LTS release.
SERVER-139192Backlog
Change plan shape counter to track batched ixscan+fetch rewrite instead of rewrite eligibility
GODRIVER-4257Needs Triage
Return the error from the Context on read/write cancellation
h3. *Context*
There are a few places in the Go Driver where we can improve the usability of Context errors:
* [connection.read|https://github.com/mongodb/mongo-go-driver/blob/e0fdd254e8517d684ae10cbf0cb7359d119ec703/x/mongo/driver/topology/connection.go#L505-L515]
* [connection.write|https://github.com/mongodb/mongo-go-driver/blob/e0fdd254e8517d684ae10cbf0cb7359d119ec703/x/mongo/driver/topology/connection.go#L435-L445]
In both places, we explicitly return {{context.Canceled}} instead of th
Components
Error Handling
SERVER-138834ClosedGone away
Complete TODO listed in SERVER-136017
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* jstests/noPassthrough/txns_retryable_writes_sessions/placement_conflict_time.js:191
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-136017&type=Code
The next steps for this ticket are to either remove the outdated TODO or follow the steps in the TODO if it is correct. If the latter, please update the summary and description of
SERVER-138833ClosedGone away
Complete TODO listed in SERVER-135300
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* src/mongo/db/index/index_access_method.cpp:445
* src/mongo/db/index/index_access_method.cpp:1020
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-135300&type=Code
The next steps for this ticket are to either remove the outdated TODO or follow the steps in the TODO if it is correct. If the latter, please update the summary and d
SERVER-138161ClosedFixed
Clean-up TenandId references in ServerParameter flow
Fixed Versions
9.1.0-rc0
SERVER-139135ClosedFixed
getESERotateActiveKEKStatus reports Date_t::min() rotation times after restart
After a restart, {{getESERotateActiveKEKStatus}} reports the persisted active KEK as {{completed}} with {{lastStartRotationTime}}/{{lastEndRotationTime}} set to {{Date\_t::min\(\)}}, because no rotation has run on the node yet. pymongo cannot decode that date \({{InvalidBSON: year 292278994 is out of range}}\), which breaks the {{ContinuousESERotation}} hook.
Fix: omit each time field until this node has run a rotation.
Fixed Versions
9.1.0-rc1025
SERVER-137810ClosedFixed
Use get_raw_key_value() in WiredTiger cursors
get_raw_key_value() is significantly faster than calling get_key() and get_value() separately.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-139112ClosedFixed
Live references to tables when a replicated drop is being applied should not be a fatal error
In the normal flow a replicated table drop should only be applied after the oldest timestamp has advanced past the drop timestamp, and so there should not be any remaining readers pinning the table. We therefore chose to make this a fatal error, as it's most likely not a transient condition and it's better to just restart a secondary than stall replication forever.
Magic Restore turns out to be an exception to this. The collection drop and table drop can be applied with no delay, and while mo
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137888ClosedFixed
Move Offline Validate code out of startup_recovery.cpp
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-139182In Code Review
Bump powercycle curator to the fixed release for CVE-2026-111497 (brotli)
h3. Problem
buildscripts/resmokelib/powercycle/setup/*init*.py downloads the curator binary from boxes.10gen.com with a hardcoded 2022-era hash (b0c3c0fc68bce26d9572796d6bed3af4a298e30e). Those binaries embed github.com/andybalholm/brotli v1.0.3, vulnerable to CVE-2026-111497 (Node2Shell brotli out-of-bounds write, CVSS 9.2, currently on CISA KEV). The binary is unpacked onto hosts being powercycled to collect system and process stats.
h3. Fix
Point the hardcoded curator_hash at 4ebe0e39c747113
Components
Build
SERVER-135203ClosedFixed
System collection contents readable through a database change stream with showSystemEvents
$changeStream's private showSystemEvents option re-admits system.js into a database- or cluster-scoped stream without requiring any privilege on system.js.
Since system.js stores server-side JavaScript that often embeds credentials, a role with only find+changeStream on a database or cluster could read it.
Fix: filter events for changeStream on system.js when showSystemEvents is used on a database- or cluster-scoped stream.
Built-in roles are unaffected.
Other system collections (buc
Fixed Versions
9.1.0-rc09.1.0-rc10258.3.169.0.78.0.377.0.48
SERVER-139169In Progress
Support errors for batching stages.
Update lookup, pre and post images and batching stages need special handling for errors. Basically add a try/catch
SERVER-137643Open
Support FieldPath and FieldRef paths in the type system
FieldRef is used for matchable paths. A $match can be on an field with an empty name such as \{$match: {"": 1}}.
FieldPath is the paths that a FieldPathExpression can reference. The path "$" is not valid. It is more restrictive and used in the pipeline module.
canPathBeArray support both, since ExpressionContext is used for all queries. One overload currently converts FieldRef to FieldPath and calls the other with it.
The schema type methods should also support both. The canPathBeArray