Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-139314ClosedFixed
Use hangInStandbyReadLoop failpoint for disagg replication pause in multikey_timestamp_transaction_large_apply_ops.js
Replace {{pauseOplogApplication}} with {{hangInStandbyReadLoop}} in the disagg path of {{{}multikey_timestamp_transaction_large_apply_ops{}}}. This pauses receiving-side processing and lets oplog entries accumulate upstream, matching the test’s intended replication pause semantics.
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-135416ClosedFixed
Modify write path (analyze) to write each new value as a new generation
Modify write path (analyze) to write gen_id as max+1
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-135016In Code Review
Implement testing for WT OTel stats
* assert that stats which gained an otel type still appear in server status section (on default and non default cursor)
* Adapt for WT stats: start a node, do I/O, confirm the metric appears with a good looking value.
* see examples {{runtime_otel_metric_creation.js}} and {{otel_metrics_file_export_helpers.js}}
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-135715In Code Review
Refactor samples calculation to conform to samples manager interface
*Refactor current analyze_cmd.cpp run methods into reusable stats refreshers*
SERVER-139081In Progress
Add Parallel Query Evergreen variants
See the test plan: [https://docs.google.com/document/d/1jai28J_fvuUnOmOQBtKOqVizgwRH7JgRj9jMEVQa2-A/edit?tab=t.0]
New variant to enable it under the cover: with background change of parallel worker threadpool size (_internalQuerySBEExchangeMaxWorkers).
SERVER-132279ClosedFixed
use after free in BlockToRow SBE stage
During stage yielding, the block from underlying stage is partially saved and owned, but the value that the current slot accessor is pointing at is not.
Fixed Versions
9.1.0-rc08.3.169.0.78.0.379.1.0-rc1026
SERVER-139540Backlog
Test the Integration with ticket granting system
We intend to verify that by running workloads containing parallelizable (including forced) queries with OLTP queries. Should verify if suggested workloads contain such queries. Parallel Query and its interfaces with the Server
https://docs.google.com/document/d/1m5Ij99riSm4FQRpc3crSG2L7E5aXPzODR1OcWlIVt6Y/edit?tab=t.0#bookmark=id.bjwi4dooncr
SERVER-139539Needs Scheduling
imlement PSAME sweep task
Task will be scheduled upon startup. Will ask all stats managers for the list of collections they hold stats from and will then cross check them vs the list of existing collections. If any stat belongs to a collection no longer existing, will schedule its deletion
SERVER-95325ClosedFixed
Remove server code related to CMConcurrency.
Leave the server param as a no-op but remove the dead code related to this server param.
Fixed Versions
8.2.0-rc0
SERVER-139538Needs Scheduling
Tidy type system headers
Anything that is in type.h and is not meant to be used externally should be removed from the header or added to a dedicated namespace. Any non-static member functions which do not access Type internals can be made free functions. Any inconsistent function names should be renamed.
This is a refactor ticket, but it an important one to do, since leaving these methods in the public API allows users to write subtly incorrect or incomplete code - such as if they used narrowField instead of the diff
SERVER-137524In Code Review
Define the PSAME feature flag and server param
Add {{featureFlagPSAMEEnabled}} (default {{{}false{}}}) to gate the whole PSAME automation engine, independently of {{{}featureFlagPersistentStats{}}}.
See WRITING-11150, Upgrade/Downgrade Requirements and Behaviors.
We also need a server parameter for it, a cluster wide one, for customers to toggle it on/off once the project has been released and the feature flag no longer exists
SERVER-139340ClosedFixed
Fix DSC mixed-version sanity check to compare Git revisions instead of version strings
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-138488In Progress
Register findAndModify query stats from the write execution path
SERVER-134366In Code Review
Parallel version of ScanStage and GenericScanStage
Clones for parallel execution would share a common structure handing off partial ranges (>> DOP) of the collection, that each stage object would consume before asking for another one
Includes unit tests for both forward and backward scanning, and performance tests to verify the speedup over the DOP=1 scenario
As parallel queries create extra Client objects that could interfere with the normal Client creation performance, create a test that measures the rate of handling new connection requ
SERVER-113403Open
Investigate test coverage for non-retriable commands triggering catalog cache
Currently, within the catalog cache we are not considering retrying on errors which are in the "retriable" error category. Based on BF-39979, we have found that tests with stepdowns and an embedded config server could fail if the primary re-election occurs immediately after [registering the term|https://github.com/10gen/mongo/blob/0ba6af3a2669b4aa3f5cacd6ccd496c0b7527a3e/src/mongo/db/global_catalog/catalog_cache/shard_server_catalog_cache_loader_impl.cpp#L556-L559], leading to [this|https://gith
NODE-7859Backlog
RTTPinger never reschedules after a failed ping or connection attempt, leaving latestRtt frozen at a stale value
h3. Use Case
h3. Expected Behavior
After a mongos restart or any failed RTT measurement, the latency measurement on a server connection should be updated at regular intervals as it was before the failure.
A failed attempt must not be recorded as an RTT sample: the pinger should discard the failed connection, leave {{latestRtt}} unchanged, and measure again after {{heartbeatFrequencyMS}}. A later successful measurement updates {{latestRtt}}.
h3. Actual Behavior and Impact
During a roll
Components
SDAM
Affected Versions
7.6.0
SERVER-134440ClosedFixed
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
NODE-7985ClosedDuplicate
RTT sampler stops after one connect or hello failure and freezes latestRtt
h3. Current behavior
RTTPinger.measureRoundTripTime returns without scheduling another sample when connect() rejects or when the dedicated hello command rejects. latestRtt stays at the previous successful sample. Streaming heartbeats use monitor.rttPinger.latestRtt, and client-side operation timeouts use the minimum RTT from these samples. The hello monitor keeps running. The sampler starts again only on monitor reset or when the server stops advertising topologyVersion.
h3. Expected behavior
A
SERVER-126409Needs Scheduling
Allow mixing of localhost and non-localhost replica set members in config
[This logic|https://github.com/mongodb/mongo/blob/4a83b9db64abedbe9de08a7602ff26724cbed4b0/src/mongo/db/repl/repl_set_config.cpp#L397-L404] current prevents this. If this isn't a capability we want to enable in general (e.g. to prevent human error) then let's add a server parameter to optionally override it.
Components
Replication