Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-135339ClosedFixed
Implement matcher narrowing with LeafArrayBehavior::kNoTraverse
The LeafArrayBehavior::kNoTraversal is required for the $jsonSchema MatchExpressions.
This ticket should implement just that case without dotted paths support (SERVER-134936).
Fixed Versions
9.1.0-rc09.1.0-rc1024
SERVER-134593ClosedDuplicate
Add basic matcher path traversal and narrowing
Type matcher::traversePath(Type input, ElementPath, Type constraint, bool assumeTrue);
Type matcher::narrowPath(Type input, ElementPath, Type constraint, bool assumeTrue);
PathMatchExpressions such as TypeMatchExpression use ElementPath to describe how they match on a match.
The ElementPath has two enums:
* LeafArrayBehavior - whether we can match within leaf array like \{$match: {x: 1}} will match 1 inside 'x' if 'x' is array.
* NonLeafArrayBehavior - whether we can traverse array
SERVER-139176In Code Review
Fix timeseries bucket min/max time metadata propagation in split aggregations
SERVER-135737ClosedFixed
Fix memory exhaustion issue in FLE2 operation stats collection
In fle_stats.cpp, each {{incrementXCount()}} function calls {{{}registerCollectionStats(){}}}, which does {{{}getOrEmplace(nss, ...){}}}. That means every operation creates a map entry if one doesn't exist yet. Both inputs come straight from the request:
- {{nss}} is the request's namespace.
- {{efc}} is parsed from the client-supplied {{{}encryptionInformation{}}}.
The only code that ever removes an entry is {{{}deregisterCollectionStats{}}}, and it runs only from the collection catalog
Fixed Versions
9.1.0-rc09.1.0-rc10258.3.169.0.78.0.37
SERVER-139183Backlog
Show parallelPlanner output in explain unconditionally once the enable knob is default on
SERVER-138609Blocked
Log hello.clientUpdate in monguard
When the server receives the hello command for the first time with the {{clientUpdate}} field, it parses and [logs it|https://github.com/10gen/mongo/blob/master/src/mongo/db/repl/replication_info.cpp#L481-L483].
Monguard should also do the same. Additionally, monguard should exclude hello commands with {{clientUpdate}} from going through the hello cache so that it is always forwarded to the upstream. We already do this for hello commands that include the {{saslSupportedMechs}} or {{compressi
SERVER-139070In Code Review
Add logging around type inference code
We need some logging since we mix multiple sources of type information and that makes it hard to track where some facts come from.
To determine a type, we:
- may read a type from the collection schema
- may run type inference on a $match
- may query path arrayness to clear arrays along a path
We should log sufficient information to reconstruct the type inference steps.
This should be debug-level logging.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-134781ClosedWon't Do
Implement clean shutdown with drain-first teardown for the pipelined oplog applier
This ticket implements clean shutdown: on the batcher shutdown signal, the dispatcher stops pulling, performs a full drain, and threads are joined in the correct order.
This ticket should include:
* The batcher shutdown signal stopping the dispatcher and issuing a full drain: queues empty, in-flight FIFO empty, lastApplied published to the last dispatched op.
* Join ordering: advancer joined before the workers.
* The drain is the batcher-driven clean-shutdown path. A global-shutdown interrupt
SERVER-139065ClosedGone away
Complete TODO listed in SERVER-137357
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* buildscripts/modules/atlas/suites/disagg_replica_sets.yml:436
* src/mongo/db/modules/atlas/src/disagg_storage/replication_coordinator_disagg_storage.h:127
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-137357&type=Code
The next steps for this ticket are to either remove the outdated TODO or follow the steps in the TODO if it
SERVER-134784ClosedFixed
Expose throughput and progress metrics for the pipelined oplog applier
This ticket exposes the throughput/progress metrics under {{{}serverStatus().repl.pipelinedApply{}}}. We will likely need to expose these via OTel as well.
This ticket should include:
* Counters: {{{}opsDispatched{}}}, {{{}opsAppliedByWorkers{}}}, {{{}opsDrained{}}}, {{{}queuedOps{}}}, {{{}queueDepthMin{}}}, {{{}queueDepthMax{}}}, {{{}lastAppliedAdvancement{}}}.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-135374ClosedWon't Fix
Enable multikey tests in the pipelined oplog applier passthroughs
The CRUD/DDL passthrough ticket (SERVER-134749) excludes multikey tests until the multikey path is ported. This ticket lifts that exclusion so the passthroughs' consistency checks exercise multikey correctness under POA.
This ticket should include:
* Lift the multikey exclusion from the POA CRUD/DDL passthroughs.
* Discovered failures filed as separate bug tickets.
SERVER-137804ClosedFixed
Conversion of Hashtable to roaring bitmap is quadratic instead of linear
Option 1) Remove chunk based migration entirely since we don't release memory until the entire object is destroyed regardless.
Option 2) Track the position of the iterator to avoid using the begin function in each iteration causing the O(n^2) time complexity.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-138554ClosedFixed
Add log ingestion rule for 136260
https://docs.google.com/document/d/1gc7Z9LLrTCLaICoFskYt_2Pz-acjJb6-S9JVekr8-i4/edit?tab=t.0
Fixed Versions
9.1.0-rc0
SERVER-134905ClosedDuplicate
Investigate why secondaries were unable to keep up with primary workload
We have had a number of escalations where resharding's cloner was apparently producing load that the replica set primary could handle, but its secondaries could not, leading to increasing replication lag over time:
* HELP-81857
* HELP-84458
* HELP-99223
In these cases, there was no obvious contention on the secondaries to explain why they were unable to keep up (at least, not obvious to someone who was not an expert in replication).
We should do a root cause analysis to determine why
Components
Replication
NODE-7890Ready for Work
Add dependabot 7 day cooldown
h3. Use Case
As a... Node Driver developer
I want... dependabot updates to have a 7 day cooldown
So that... a newly published malicious or unstable package will not be picked up for 7 days, allowing for maintainers to review and fix a malicious or unstable package
h3. User Experience
* for all repos: dependabot updates come in a week later than usual
* this must be done for all of our repos: driver, bson, conn string, etc.
h3. Dependencies
* none
h3. Risks/Unknowns
* an impor
NODE-7720Ready for Work
Pin GitHub Actions to commit SHAs across Node packages
Semgrep flagged two mutable tags on [mongodb-client-encryption#137|https://github.com/mongodb-js/mongodb-client-encryption/pull/137#discussion_r3659322472]; I deferred both with `/other` pointing here.
h3. Use Case
As a maintainer of the Node packages
I want every `uses:` reference pinned to an immutable commit SHA
So a repointed upstream tag cannot change what runs in our release jobs without a diff in our repos.
`vN` tags move by design on every patch, and that same git write is the
Components
CI
SERVER-137120ClosedFixed
Encode query settings into join plan cache key
The classic plan cache key [encodes query settings|https://github.com/10gen/mongo/blob/f5541773c70ff776becf870803ca2835a58f08af/src/mongo/db/query/plan_cache/plan_cache_key_info.h#L47] so that the same shaped query run under a different configuration of query settings will not hit the cache entry. We should do a similar encoding for the join plan cache key.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-130261ClosedFixed
dropAllRolesFromDatabase uses malformed cleanup predicates and can leave stale role references that later rebind to recreated roles
h1. Security Context
See SECBUG-1278 for problem description, security impact, preconditions, reproduction steps, and severity rationale.
h1. Root Cause
* src/mongo/db/commands/user_management_commands.cpp: the cleanup filters in dropAllRolesFromDBTxnFn are built as nested plain objects that do not match array elements:
** User-doc cleanup uses filter rolesMatch = \{roles: {db: X}} — this matches a roles field whose value is the document \{"db": X} exactly, not an array element containin
Fixed Versions
9.1.0-rc09.1.0-rc10258.3.169.0.78.0.377.0.48
SERVER-130169ClosedFixed
X.509 cluster-membership separation check can fail open on user lookup errors and grant internal authorization
Related to CVE-690 Engineering Work and SECBUG-1282
h1. Security Context
See SECBUG-1282 for problem description, security impact, affected conditions, reproduction steps, and severity rationale.
h1. Root Cause
* src/mongo/db/commands/authentication_commands.cpp: user-existence check collapses all non-OK results into "user absent" via .isOK(); ErrorCodes::UserNotFound and transient failures (MaxTimeMSExpired, NotPrimary, lock timeout, Interrupted) are indistinguishable to the caller
* s
Fixed Versions
9.1.0-rc09.1.0-rc10258.3.169.0.78.0.377.0.48
SERVER-130496ClosedFixed
Fix integer overflow in FLE2 readTags accumulator
h1. Security Context
See SECBUG-1461 for problem description, security impact, preconditions, reproduction steps, and severity rationale. A verified Docker-based PoC is included in that ticket.
h1. Root Cause
The bug is a type-width mismatch in src/mongo/crypto/fle_tags.cpp::readTags(). Three lines interact to produce the exploit:
+*Line 148 — accumulator declared as uint32_t*+
FLEEdgeCountInfo::count (fle_crypto_types.h:92) is uint64_t. The accumulation loop silently truncates each
Components
Queryable Encryption
Fixed Versions
9.1.0-rc09.1.0-rc1025