Known Issues
5166 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-139080Backlog
Investigate the monguard FindFilterSort tail regression
SERVER-137897 found one workload shape (FindFilterSort in top_ten_queries_locust) where monguard costs more than noise with socket handoff. It measured these latencies:
||Arm||p95||p50||avg||
|No monguard|3.01ms|1.81ms|2.33ms|
|Monguard proxy|5.91ms|2.71ms|3.70ms|
|Monguard handoff|4.89ms|2.24ms|3.08ms|
Handoff recovers part of the proxy regression. Handoff p95 latency is 1.9ms (~60%) higher than no-monguard (n=1). The original SERVER-137897 report calculated the regression as +37% p95
SERVER-139340In Code Review
Fix DSC mixed-version sanity check to compare Git revisions instead of version strings
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-138125ClosedDuplicate
Membership-based rid honoring in oplog application
The applier honors {{rids}} iff the node holds a {{groupId}}, replacing the per-collection catalog check. This is the core of the feature: in-group nodes apply updates and deletes by {{rid}}, out-of-group nodes fall back to {{_id}} lookup.
h2. Details
- Relax [tassert 11454701|https://github.com/10gen/mongo/blob/5fce3b99dea39b4f9453477819cedd935fba9a2c/src/mongo/db/repl/oplog.cpp#L2590] ({{rid}} presence iff collection option) to a membership-plus-in-scope check.
- In-group: inserts at the
SERVER-110207ClosedFixed
Block shutdown during endNonBlockingBackup()
It's possible for the storage engine to shut down during this call, and this function calls into the storage engine to close the backup session.
This will likely have a similar approach to SERVER-99010
Fixed Versions
8.3.0-rc08.0.39
SERVER-124059ClosedFixed
fix bazel wrapper nested python installs
some of our wrapper target run nested multiple bazel commands, causing the intermittent issues in the way python is detecting the current modules in the python path.
common failure mode would be something like:
Traceback (most recent call last):
File "/data/OTHER_HOST/mongo/bazel/wrapper_hook/wrapper_hook.py", line 193, in <module>
main()
File "/data/OTHER_HOST/mongo/bazel/wrapper_hook/wrapper_hook.py", line 86, in main
from bazel.auto_header.auto_header import gen_auto_hea
Fixed Versions
9.0.0-rc08.3.18
SERVER-83322ClosedFixed
Compound Wildcard Index bounds reverts to [MinKey, MaxKey] when filter contains fields outside CWI
I'm observing a weird behaviour with CWI's, where queries containing filters beyond those covered by the CWI, results in the `indexBounds` to revert to '[MinKey, MaxKey]`.
Say I have CWI:
If I query only using fields contained in the CWI:
I get this winning query plan:
But if I add an extra criteria, say `{ Foo: 'Bar' }`:
I suddenly get a query plan where the index bounds on `CustomFields.Prop` and `Date` are both `[MinKey, MaxKey]`:
{code:js}
{
"stage": "FETCH",
"
Affected Versions
7.0.18.0.0-rc0
Fixed Versions
9.0.0-rc08.2.138.3.78.0.39
SERVER-127745ClosedFixed
Resharding recipient service deadlock on rapid stepdown then stepup
A resharding recipient node can permanently deadlock during step-up: it wins the election and reports PRIMARY, but never becomes a writable primary. The node logs 8025900 ("onStepUpComplete stepping up all services") but never 21331 ("Transition to primary complete"), and stays in this state until the process is killed. Reproduced on v8.0 and master; observed in production in HELP-93835 (v8.0.19).
h3. How the deadlock forms
It takes two failovers of the same node while the resharding recipie
Components
Index Builds
Fixed Versions
9.1.0-rc09.0.48.0.378.3.18
SERVER-132822ClosedFixed
createIndexes reports success without creating the index when a concurrent moveCollection replaces the collection
{{createIndexes}} can reply {{ok: 1}} without having created the requested index. The client is
told the index exists, and any later operation on it fails with {{IndexNotFound}} for example.
h2. How it happens
# {{createIndexes}} resolves the target collection UUID and releases all locks, then blocks in the
{{maxNumActiveUserIndexBuilds}} queue in {{IndexBuildsCoordinatorMongod::_startIndexBuild()}}. The
config fuzzer had set the knob to 1, so the wait lasted ~15 seconds.
# A concurrent
Fixed Versions
9.1.0-rc09.1.0-rc10229.0.9
SERVER-134723ClosedFixed
Fix crash if sync source resolver fails to find sync source for catchup
There seems to be a crash coming from [this log message|https://github.com/mongodb/mongo/blob/v8.0/src/mongo/db/repl/bgsync.cpp#L396]. This is logged when the sync source resolver can't [find an appropriate sync source|https://github.com/mongodb/mongo/blob/v8.0/src/mongo/db/repl/bgsync.cpp#L389]. However, {{getSyncSource()}} requires that there be a valid sync source stored in the response object and that the [return status is OK|https://github.com/mongodb/mongo/blob/v8.0/src/mongo/db/repl/sync_
Components
Replication
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.188.0.39
SERVER-134440In Code Review
Add logging for log ingestion when sharding metadata indexes are created on non-empty collections during step up
SERVER-132525 removed a self healing behavior when sharding metadata indexes were found to be missing during step up. The only known case where this should happen currently is in DSC since there were 9.0 DSC clusters running before the 9.0 upgrade procedure was complete, thus allowing these clusters to bypass the setFCV protocol. However, other than this, there should be no cases in which this path is taken.
We would like to add logging and a log ingestion rule in 9.0 to validate this and mak
Fixed Versions
9.0.9
SERVER-134555ClosedFixed
Translate queries on an empty-string metaField instead of requiring a non-empty metaField
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.188.0.39
SERVER-135202ClosedFixed
Make gRPC concurrent connection unit tests use a larger timeout
Please refer to [this comment|https://jira.mongodb.org/browse/BF-46196?focusedCommentId=8915212&focusedId=8915212&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-8915212] in BF-46196 for context.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.18
SERVER-134991ClosedFixed
Add completion log for chunk migration
Similar to [resharding|https://github.com/mongodb/mongo/blob/6d990baca59ce7cdddded93409e0266d1ac6a45a/src/mongo/db/s/resharding/resharding_coordinator.inl#L2750].
Similar to resharding, we want the log to only appear once per movechunk execution (no duplicate logs when movechunk is re-run during recovery). And potentially capture all errors encountered during moveChunk (should at least have covered cases when the countDonorMoveChunkAborted stat is incremented).
Note that in the resharding lo
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.0.39
SERVER-134972ClosedFixed
Improve log message when failing to apply batch of oplog entries
[This log message|https://github.com/10gen/mongo/blob/94af1ff2554cd464c9e24a16f26acf4cbcb6b990/src/mongo/db/repl/initial_sync/initial_syncer.cpp#L2048] is logged whenever initial sync fails to apply a batch of oplog entries. However, it doesn't contain information about the batch, which makes it difficult to debug any incidents. We should add some additional information to this message.
Components
Replication
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.188.0.39
SERVER-135249ClosedFixed
Prevent change streams metrics exceptions from blocking serverStatus completion
Inside the generation code [here|https://github.com/10gen/mongo/blob/f3f8c0277a2cbd06da47c7b3d33bcc8161dd581e/src/mongo/db/stats/change_stream_preimages_server_status.cpp#L68-L69] accessing the storage layer can lead to a max WiredTiger sessions error if we're at capacity. This layer does not handle this error, and the exception will propagate up to the serverStatus collector and prevent a particular serverStatus call.
This error can be seen in execution 3 of [this task|https://spruce.corp.mo
Components
Change streams
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.188.0.39
SERVER-135720ClosedFixed
Disable mongor tests
These tests are generating BFs but are testing a feature which is no longer under development or supported.
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.18
SERVER-136131ClosedFixed
Fail fast DDL operations if there is an index build running on affected namespaces
Right now the DDL operations does not respect running index builds.
This can lead to availability issues like in HELP-97655 and HELP-97212.
Some custom error handling is already applied to handle this edge case in some DDL coordinator (like SERVER-117624), but there is no generic solution implemented yet.
This task is about adding a generic best effort check for DDL operations to avoid running during an active index build on the affected namespaces.
* The solution should be a best effort che
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-135803ClosedFixed
Fix UnorderedTicketSemaphore timing out after a permit becomes available
{{UnorderedTicketSemaphore::acquire}} ignored the boolean result from {{{}_permits.waitUntil{}}}. With a short explicit deadline, a waiter awakened by a newly released permit could be mistaken for a deadline expiration because {{{}deadline == until{}}}. This could return a false timeout even though a ticket was available
As part of this ticket we should check the {{waitUntil}} result and enter the timeout path only when {{!waitResult && deadline == until.}}
Affected Versions
8.3.0-rc09.0.0-rc09.1.0-rc0
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.78.3.18
SERVER-136697ClosedFixed
Retry change stream FSM DDL commands on concurrent index builds
The change stream sharding FSM tests run two writers against the same database (`test_cs`) with different collections. The writers run independent command streams, so one writer can start an index build (`createIndexes` / `shardCollection`) on its collection while the other issues a database-level DDL command. In particular, a `movePrimary` on test_cs races with a sibling writer's index build: the `movePrimary` coordinator scans the database for in-progress index builds and rejects the command w
Components
Change streamsTesting Infrastructure
Fixed Versions
9.1.0-rc09.1.0-rc10249.0.9
SERVER-136811ClosedFixed
Update external auth tests with new LDAP test endpoint
The old test host {{ldaptest.10gen.cc}} is scheduled to be taken down on October 26, 2026. The external auth tests that depend on this hostname must be updated to {{ldaptest.build.10gen.cc}} , otherwise they will fail.
Fixed Versions
9.1.0-rc09.1.0-rc10255.0.366.0.314.4.337.0.499.0.98.3.188.0.39