Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-119036In Code Review
Delete old get_executor
When this project is complete and we've successfully released the new get executor, we can delete the old one.
This includes all of the old planner interface classes that target a specific engine
Fixed Versions
9.1.0-rc09.1.0-rc1024
SERVER-136257ClosedFixed
Disambiguate null/missing within streaming $group keys
Fixed Versions
9.1.0-rc09.1.0-rc1025
NODE-7735ClosedDone
[Mastra] Hybrid Search support
Mastra RAG is still vector-only, and this issue proposes hybrid retrieval that combines semantic and lexical matching. For developers, better retrieval quality means fewer custom ranking hacks; for end users, it means more accurate answers; for MongoDB, it strengthens the case that MongoDB can power not just storage but high-quality RAG experiences inside Mastra
Components
ABX
SERVER-138323ClosedFixed
Turn off replicated size and count for initial_sync_preserves_initiating_set_marker.js
The test expects specific oplog entries during the initial sync fetching while replicated size and count changed the logic of oplog fetching.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-138082ClosedFixed
Extend collation testing for lookup result cache
Just adding one more test case for collation! Where we have different local/foreign collations and then override the collation also
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-133545ClosedFixed
renameCollection with dropTarget can lead to orphaned replicated size-count metadata entries
A {{renameCollection}} with {{dropTarget}} destroys the collection occupying the target namespace, but produces no {{drop}} oplog entry for it. The destruction is expressed only as a field inside the rename entry.
This means that we can have orphaned metadata store entries for the collections that are dropped with {{{}dropTarget{}}}.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137427ClosedFixed
Enable Record Store Parallelism in Offline Validate
Enable the parallelism feature flag and change the defaults so offline validate runs in parallel.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-101360ClosedFixed
Delete src/mongo/crypto/fle_options.cpp
This is currently not in a Bazel target. Please either delete it or move it into a Bazel target so it can be linted for errors.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137833ClosedFixed
Complete TODO listed in SERVER-101360
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* bazel/wrapper_hook/lint.py:378
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-101360&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 do
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137483ClosedFixed
Generalize LiteParserRegistration to allow for binary-compatible fflags
Currently, only stages passing IFR flags are allowed to register fallback parsers. Add a branch that can check a binary-compatible feature flag at parser selection time and remove any logic gating fallback parser registration on IFR flags.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-134448ClosedFixed
Expand landlock filesystem policy grants
Expand the Landlock filesystem rules to add permissions on the following paths:
* extensionsConfigPath
* query extension test_examples directory
* query extensions .so files specified in each query extension's ".conf" YAML file, and its signature ".so.sig" file.
* the current working directory (so core dumps can be written)
* OTEL paths (set parameters: openTelemetryMetricsDirectory, opentelemetryTraceDirectory, openTelemetryPrometheusMetricsDirectory, openTelemetryPrometheusMetricsPat
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-138088ClosedFixed
UBSan negation of INT_MIN in ordered_ticket_semaphore_fuzz_test FuzzPriorityOrdering
h2. Problem
{{FuzzPriorityOrdering}} in {{src/mongo/db/admission/ticketing/ordered_ticket_semaphore_fuzz_test.cpp}} normalizes each fuzzed {{int32_t}} admission count with {{c = std::abs(c) % 1000}} (line 42). The domain is {{fuzztest::VectorOf(fuzztest::Arbitrary<int32_t>()).WithMinSize(1).WithMaxSize(16)}} (lines 136-137), which includes {{-2147483648}} ({{INT_MIN}}). {{std::abs}} on a signed {{int}} of that value is undefined: the negation cannot be represented. UBSan aborts the test and {{li
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-136942ClosedFixed
Fix off-by-one errors in time-series bucket validation
Change the bound checks to {{idx >= maxCount}} and {{{}idx >= bucketCount{}}}.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-135413ClosedFixed
Introduce gen_id into the stats _id schema; always write gen_id = 0 for now
Modify the schema to introduce this new value into the _id, defaulting to 0 and the total pages into the body.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137784ClosedFixed
Add awaitReplication calls to auth_no_pri.js
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137767ClosedFixed
Decouple jstestfuzz --branch from branch_name so non-version branches (e.g. main-staging-poc) work
h2. Overview
All jstestfuzz\-driven tasks fail immediately in the _mongodb\-main\-staging\-poc_ Evergreen project \(a branch closely tracking master\). This is not a real test failure — the fuzzer never runs, because jstestfuzz rejects the branch name before generating any tests.
h2. Root cause
The Evergreen built\-in expansion {{branch\_name}} is passed straight through to jstestfuzz as its {{\-\-branch}} argument, in two places:
* {{evergreen/jstestfuzz\_run.sh}} \(final line\): {{./src/sc
Fixed Versions
9.1.0-rc0
SERVER-132395ClosedFixed
Exclude findAndModify_update_grow from disagg_concurrency_simultaneous
See SERVER-132391 for the follow-up
Components
Concurrency
Fixed Versions
9.1.0-rc0
SERVER-137972ClosedDone
Operations are not killed on client disconnect when a TLS client sends close_notify before closing
*aggregate* opts into kill-on-disconnect ({*}handleMarkKillOnClientDisconnect{*}),
and over plain TCP an aggregate whose client has disconnected is killed with
*ClientDisconnect* within about a second. Over TLS, if the client closes cleanly
(sends a `close_notify` alert, then closes the socket), the operation is never
detected as disconnected and runs to completion. The Go driver always closes
this way, because Go's *crypto/tls* sends *close_notify* on {*}Close(){*}.
The likely cause is
Affected Versions
8.3.117.0.45
SERVER-137867ClosedFixed
Swap writeWatermark() and flush() declaration order
writeWatermark() and flush() are defined in different orders in size_count_checkpoint_flusher.\{h, cpp}. Make the order consistent.
Fixed Versions
9.1.0-rc09.1.0-rc1025
SERVER-137861ClosedFixed
Remove stale comment in size_count_checkpoint_flusher.cpp
Remove stale comment from commit 8d836f101af7 ("SERVER-121018 Introduce tailable size count checkpointing (#55009)"). There is no {{OplogScanResult::startAfter}} field anymore, and the previous valid-as-of timestamp is not always the source of truth for the next checkpoint.
Fixed Versions
9.1.0-rc09.1.0-rc1025