Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-123814ClosedFixed
Update team tag in resmoke tasks.yaml
"workload_scheduling" is still used in [https://github.com/10gen/mongo/blob/master/etc/evergreen_yml_components/tasks/resmoke/server_divisions/clusters_and_integrations/tasks.yml#L609]
We should replace it with "workload_resilience"
Fixed Versions
9.1.0-rc09.1.0-rc1025
NODE-7895Backlog
URI values override programmatic options
h3. Expected Behavior
As a developer passing both a connection string and a MongoClientOptions object,
I expect the programmatic value to win for every option, as documented on MongoClient ("The programmatically provided options take precedence over the URI options.")
When the same setting is provided in both places, either under the same option name or under different option names that configure the same thing.
h3. Actual Behavior and Impact
* parseOptions collects the programmatic value
SERVER-130179ClosedFixed
Propagate cost information to explain in sharded environment and differed engine choice
When the cbr_shard_filter.js is run with deferred engine choice, it passes an empty EstimateMap to the plan executor, so costEstimate is absent from explain output.
Fixed Versions
9.1.0-rc09.1.0-rc1024
PYTHON-6145ClosedFixed
Cursor and AsyncCursor annotate `let` as bool although runtime requires a document
PyMongo 4.18.1 and 4.18.2 / CPython 3.14.6 / mypy 2.3.1. No MongoDB server is needed.
Both native cursor constructors declare `let: Optional[bool] = None`. Their implementations call `validate_is_document_type("let", let)` for any non-None value. A document constructs successfully; a boolean raises TypeError at construction. Mypy reports two arg-type errors for document arguments and accepts the boolean arguments.
The affected released source is [Cursor](https://github.com/mongodb/mongo-py
Fixed Versions
4.20.0
SERVER-119076ClosedFixed
Add query stats counters for common plan shapes in new get_executor path
We'd like to reuse our QSN analysis functionality built in SPM-4435 to track how often common plan shapes are used. This will help inform further releases. If we choose to enable SBE for more plans, we'll have an idea of how many queries will be exposed to the new code paths.
Fixed Versions
9.0.0-rc1
SERVER-130223ClosedFixed
Add query stats plan shape counters for QSN types and access paths
Fixed Versions
9.0.0-rc1
SERVER-130432ClosedFixed
Move plan shape counter logic into its own folder
Fixed Versions
9.0.0-rc1
SERVER-132297ClosedFixed
Prevent plan shape counters from executing on foreign side of classic $lookup
In the current implementation of the plan shape counters, we increment for each query to the foreign collection when a $lookup is executed in classic. Our intention was to only increment them for the local side, this fixes the issue by first checking some data on the canonicalQuery about if we're in a $lookup or a subpipeline
The counters are incremented for every query to the foreign collection, which has caused minor perf regressions
Fixed Versions
9.1.0-rc09.0.0-rc2
SERVER-137181ClosedFixed
Improve sample reading for Histograms build
Fixed Versions
9.1.0-rc09.1.0-rc1025
PYTHON-5315ClosedDone
Test Failure (Windows) - test.asynchronous.test_srv_polling.TestSrvPolling
h3. *Name of Failure:*
test.asynchronous.test_srv_polling.TestSrvPolling.test_recover_from_initially_empty_seedlist, test.asynchronous.test_srv_polling.TestSrvPolling.test_recover_from_initially_erroring_seedlist
h3. *Link to task:*
https://spruce.mongodb.com/task/mongo_python_driver_test_win64_test_v5.0_python3.13_async_noauth_nossl_standalone_patch_3723edc199a45e9dd0012ce449671eddedae40ee_67fd2d0bb41c2300078dd21d_25_04_14_15_43_19/tests?execution=0&sortBy=STATUS&sortDir=ASC
h3. *Cont
SERVER-137942ClosedFixed
Make StorageEngineMock throw to test DBException handling
{color:#c1c7d0}In the ovverride of tryPauseAutoCompactForReplicaSetWritesBlock, the StorageEngineMoch returns an error status rather than throwing, so the new test does not exercise the observer helper's {{DBException}} catch. Removing that catch would still pass the test, although an escaping exception would terminate the process through {{RecoveryUnit::Change::commit() noexcept.}}{color}
{color:#c1c7d0}As part of this ticket, we should make this mock throw so the existing commit and write-b
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-93113ClosedFixed
Implement ExhaustResponseReader
ExhaustResponseReader will be a new type returned from NetworkInterface::startExhaustCommand that will allow the controller to dictate when exhaust responses are read from the underlying connection, rather than having the NITL reactor thread do that work. In this ticket, we should implement the type according to the design and add C++ unit or integration tests for it as a component. Integrating it with NITL and ThreadPoolTaskExecutor will happen in a subsequent ticket.
Fixed Versions
8.1.0-rc0
SERVER-79871ClosedFixed
Simplify OperationId
[OperationId|https://github.com/mongodb/mongo/blob/master/src/mongo/db/operation_id.h] looks much more complex than necessary, it could be an atomic 64 integer instead. As far as I can tell we only require uniqueness, we never check the registry outside of testing code.
Fixed Versions
7.3.0-rc0
SERVER-137287ClosedFixed
Skip move range coordinator warning assertion in last-patch multiversion suites
{{chunk_migration_does_not_block_indefinitely_acquiring_strong_locks.js}} [coordinator-warning assertion|https://github.com/10gen/mongo/blob/d51d25a14a958dc1fbfbae477eb405f4b8b8c4e8/jstests/sharding/chunk_migration_does_not_block_indefinitely_acquiring_strong_locks.js#L82-L85] fails in the 9.0 {{sharding_last_patch}} suite when the donor runs 9.0.4. In particular, the migration correctly returns {{{}LockTimeout{}}}, and writes complete successfully. However, 9.0.4 logs the coordinator failure as
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-136153ClosedFixed
Unify canPathBeArray handling in the graph to use type information
We have a few divergent code paths which are used to determine if a field in question is array.
The public canPathBeArray method on the graph is the most complete implementation. But internally, when building the graph, we use different more limited checks.
For example, in a rename \{$set: {a: '$b.c'}}, if 'b.c' is a constant, we will not apply the walkConstant logic if 'b' has a constant value. Whereas if we query "b.c" using the public API we will.
After we unify the code, the followi
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137530ClosedFixed
Type information in the graph should flow through renames
The following should work:
[\{$match: {$and: [{x: {$type: 'number'}},
\{x: {$not: {$type: 'array'}}}]}},
{$set: \{a: '$x'}}]
The type of 'x' is number. We should report that the type of 'a' is number too!
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-136868ClosedFixed
resetPlacementHistory can leave config.placementHistory recreated without its secondary indexes by a concurrent DDL commit, crashing the config server on step-up (fassert 40184)
During resetPlacementHistory InitializePlacementHistoryCoordinator drops config.placementHistory in its kDefineInitializationTime phase, but only re-creates the collection and its two secondary indexes (nss_1_timestamp_-1 unique, timestamp_-1_nss_1) two phases later, in kComputeInitialization. In between, DDLs are unblocked: the kUnblockDDLs cluster-parameter broadcast takes effect per-shard as soon as each shard receives it, even while the phase handler itself is still running.
A DDL that co
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-135776ClosedDuplicate
Skip config index rebuild if ther this one in progress carried across step up
As concluded from [BF-46384|https://jira.mongodb.org/browse/BF-46384] it's possible that the if there is a config index being built in progress, it still continues to be building even if the node loses the election and steps down. In such cases, it's possible that it can go through a step up and currently it would lead to attempt to create another config index that would be conflicting with the one that is in progress, which leads to a fatal failure. Ideally instead we should simply skip re-buil
CSHARP-6116ClosedDone
Trim the MONGODB-AWS auth test matrix
The MONGODB-AWS auth tests ran every credential scenario against every server version on Windows, Ubuntu 20.04 and macOS: 20 tasks per commit, all on {{net6.0}}. The driver's AWS auth code doesn't depend on server version, and the auth spec sets no version requirement. The matrix also never tested the AWS SDK builds that .NET Framework and .NET 8+ applications load.
Trimming the matrix to one 4.4 floor check plus each scenario on {{latest}}, spread across {{net472}}, {{net6.0}} and {{net10.0}
Fixed Versions
4.0.03.next