Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-110439ClosedWon't Do
Unify spilling metrics reported in serverStatus
All spilling stages should report 4 metrics to serverStatus
spills : number of times the stage spilled,
spilledBytes : number of spilled bytes,
spilledRecords : number of spilled records,
spilledDataStorageSize : size of spilled data on disk
From all stages that spill, Spool (classic) does not report any metrics in ServerStatus while Sort reports spillToDisk and spillToDiskBytes. We should fix those two stages to report the same metrics as the rest of the spilling stages.
SERVER-130246ClosedFixed
Enforce strict: false on IDL reply structs to prevent mixed-version breakage
h3. Background
When a new field is added to a command reply via {{generic_argument.idl}} or a dedicated reply IDL struct, an older mongos (or shard acting as sub-router) parses the reply from a newer binary using a strict IDL struct ({{{}strict: true{}}}), and rejects the unknown field with {{{}IDLUnknownField{}}}. This cannot be gated behind FCV because FCV only tracks major.minor, not patch versions.
This issue was surfaced in SERVER-128128 and discussed in #server-dev-platform. The cons
Fixed Versions
9.1.0-rc0
SERVER-139330ClosedFixed
Remove exclusion of secondary_reads_waitForShardingInit.js from no_passthrough_disagg_override_sharded
The change in -SERVER-138952- was intended to adapt and re-enable {{jstests/sharding/secondary_reads_waitForShardingInit.js}} in the {{no_passthrough_disagg_override_sharded}} suite. However, the merge with master did not apply correctly, and the test remains excluded.
As part of this ticket, we should remove the exclusion so the test runs as part of {{{}no_passthrough_disagg_override_sharded{}}}.
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-136285ClosedFixed
Possibly-dangling collator pointer after ExpressionSortArray clone
Prior to SERVER-119491 (added in 9.0), we had this cloneUsingNewExpCtx logic which used to serialise and reparse an expression tree using a new ExpressionContext. This re-creates the expression and therefore could not have pointers dangling to the previous expression context.
After SERVER-119491, the clone method of all expressions was updated to accept a target ExpressionContext* for the cloned result.
In clone(), ExpressionSortArray, ExpressionTopN, ExpressionTop, ExpressionBotton and Ex
Fixed Versions
9.1.0-rc09.0.79.1.0-rc1026
SERVER-137866ClosedFixed
Set sample size per collection with collMod
See [https://docs.google.com/document/d/1SKkL4sS1CpmgOLSU9UZbAQ14-hrz2eHCvYXoCL6zQzg/edit?tab=t.0#heading=h.2s9zdv6a57gn] (in the Examples section) for some ideas of what the structure could be.
Keep SPM-2924's collMod needs (see section in [technical design|https://docs.google.com/document/d/1RqqoEMx2aHRf0WRbGy-4AVb3NN-87hd9V7whIA-Y4yA/edit?tab=t.t60zi2ksw6d#heading=h.rfgfbankw2v9]) in mind as well so our attributes can coexist nicely.
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-137792ClosedFixed
Add the rangeDeleterThreadsCount server parameter.
Introduce the {{rangeDeleterThreadsCount}} server parameter to configure the maximum number of worker threads used by the range deleter.
The parameter defaults to {{{}1{}}}, can be set at startup or changed at runtime and supports a maximum of {{256}} worker threads.
The following FTDC metrics should also be added:
* {{{}numRunningRangeDeletionTasks{}}}: the number of range-deletion tasks currently running.
* {{{}maxRangeDeleterThreadPoolSize{}}}: the configured maximum size of the ran
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-139427In Code Review
Improve multi-range clustered collscan bounds extraction performance
Extracting bounds for eg. a long $in list slowed down quite a low after SERVER-125913, hurting query performance in cases where the exec gains are low.
SERVER-139156ClosedFixed
Fail change stream on overly large transactions
Make a change stream fail with {{ChangeStreamFatalError}} when encountering a transaction with INT32_MAX or more operations inside the same transaction.
The {{txnOpIndex}} value in this case would exceed INT32_MAX and cannot be safely encoded.
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-135321ClosedFixed
Track sealed-segment materialization thru log service
PALI currently relies on CMS polling to determine whether sealed segments are fully materialized. This provides only completion information and does not expose incremental progress.
Use log service readLog streams to monitor incomplete sealed segments and publish granular cell-and-segment materialization updates to the materialization tracker
Retain CMS-based initialization thru getLogSegmentList. During runtime, open monitoring streams as segments are sealed and release them wh
Fixed Versions
9.1.0-rc1025
SERVER-137507Open
Complete TODO listed in SERVER-121845
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* jstests/disk/wt_size_storer_cleanup_replica_set.js:60
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-121845&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 this ticket to represent the
SERVER-101414ClosedFixed
Documentation Updates
This ticket is to update all types of documentation API Guide,
Programming Guide, and Architecture Guide related to the project.
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-106835Open
Hanging drop-pending collections without holders in MongoDB 8.x
In MongoDB version 8.x, there is an issue where some collections marked for deletion (in a "drop-pending" state) remain hanging even after all operations involving them have ended—that is, even without any active holders. These collections are not properly deleted and thus block resource cleanup, which can lead to degraded performance and other errors (e.g., {{{}ObjectIsBusy{}}}). The issue is rare but confirmed on multiple instances.
Affected Versions
8.0.9
SERVER-112803Backlog
Complete TODO listed in SERVER-87119
There is a TODO in the codebase referencing a resolved ticket which is assigned to you.
Please follow this link to see the lines of code referencing this resolved ticket:
https://github.com/10gen/mongo/search?q=SERVER-87119&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 this ticket to represent the work you're actually doing.
SERVER-137370In Code Review
Read from ORP in stages and resume from last received entry
Right now we ask ORP for an LSN range in a single request so the request can be large ranges and the stream can stay open for a long time. ORP wants to be able to break up long reads for load shedding purposes
On our end, we'd need to split ORP reads into stages reads and have the standby continue from the last oplog entry it received. these planned disruptions from ORP should come with a marker (like complete = False) so we know that it's not a failure and it should not cause a backoff or co
SERVER-137936Open
To have a single configuration parameter to set WiredTiger eviction workers at runtime
WT Eviction workers can be set runtime as (thread_min, thread_max).
WT has internal logic to manage the eviction workers as a pool of threads and dynamically adjust optimal eviction workers. This feature is not used as it was not giving consistent results.
Customers will always use the same number for both, thread_min and thread_max.
A single configuration parameter to set both the thread_min and thread_max.
SERVER-115109Backlog
add originalQueryShapeHash to getMore and explain slow query logs on mongod
SERVER-107427 added originalQueryShapeHash to aggregate, find, count, and distinct commands slow query logs on mongod. The router forwards the query shape hash calculated on the router to the shards. This allows query stats on mongos to be connected to slow query logs on mongod. See SERVER-107427 for more details
SERVER-135094ClosedFixed
transaction_aborted_under_cache_pressure.js flakes on master because WT-18210 shrinks the drain-loop cache-pressure window
h2. Summary
{{jstests/noPassthrough/txns_cache_errors/transaction_aborted_under_cache_pressure.js}} intermittently fails on {{mongodb-mongo-master}} at:
The failure has been observed in BFG-3557872, BFG-3611646, and BFG-3638297, all on {{{}amazon-linux2023-arm64-static-compile{}}}. These failures were linked under BF-46354, but appear to be a separate, lower-rate master flake rather than the v9.0 backport-gap failure addressed by SERVER-132736.
h2. Background
The test relies on a
Fixed Versions
9.1.0-rc09.1.0-rc1024
SERVER-139428Open
Make LSN progress check stricter to avoid unbounded retries
This behaviour predates SERVER-137175 (SECBUG-6458 even mentions this). SERVER-137175 change maintained the existing magic restore retry policy and added a stricter optional LSN progress check. The unbounded retry originates in SERVER-129948. I'm not sure that this should be considered a security issue (the test uses a fake ORP stub, and this means that for this to be a security vulnerability the attacker would have to be able to control the ORP service), but it is a latent and possible issue. W
SERVER-139062Open
Complete TODO listed in SERVER-132559
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* src/mongo/db/modules/atlas/src/disagg_storage/pali/log_server_manager.cpp:414
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-132559&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 this
SERVER-138374ClosedWon't Fix
Raise workload client CPU threshold for PALI chaos tests
PALI chaos tests can fail with CPUUsageError when workload client CPU usage exceeds the current 90% threshold. Failover events can cause expected CPU spikes from reconnects and retry storms and end up with false test failures like BF-46424
Raise the CPU threshold in configurations/test_control/test_control.pali_chaos_locust.yml