Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-136795Open
Use the status returning URI parser
Replace calls to parseServerURI with parseServerURIStatus so invalid URIs return an error instead of throwing. Handle the error or fail with an invariant check (in PALI code) to avoid uncaught exceptions
SERVER-136794ClosedWon't Fix
Use common CMS retry and timeout helpers
Move CMS endpoint rotation and timeout handling and any other shared logic into helpers so calls such as GetLogSegmentList and resolveORPEndpoint have consistent behaviour as RPCs that we send through to CMS.
SERVER-135623Open
Sync wiredTigerExtraDiagnostics with WiredTiger options
The MongoDB wiredTigerExtraDiagnostics validator is out of sync with WiredTiger’s current extra_diagnostics options.
SPM-2897 added wiredtiger asserts to mongodb from 7.0. However, Wiredtiger has renamed their options in the import [713b8810e0f1|https://github.com/10gen/mongo/commit/713b8810e0f1] as part of WT-10439 Update diagnostic categories.
Valid WiredTiger options now are rejected by MongoDB.
WiredTiger valid extra_diagnostics:
Server's WT global options:
We should updat
SERVER-133343Open
Investigate PALI rate limiter perf test benchmarks
right now the PALI rate limiter perf tests run on the waterfall weekly. there are no existing benchmarks to detect perf regression etc.
SERVER-137257Open
Avoid copying oplog/phylog through encryption
The oplog and phylog payloads are copied several times between the WiredTiger record and the log service. Five of these copies are unnecessary.
On the primary, three:
# Prepending the 2-byte compression header. The uncompressed path allocates a buffer of header + payload and copies both into it, so the encryptor is handed one contiguous plaintext.
# The ciphertext out of the encryptor. The encryptor returns the ciphertext in a std::vector, and the caller copies it into a local before use.
SERVER-134828Open
Print all thread stacks blocking in shutdownEnforcer.cpp
[`src/mongo/db/modules/atlas/src/disagg_storage/pali/util/shutdown_enforcer.cpp:41-48`|https://github.com/10gen/mongo/blame/fca7dfc5c14483db09dfc25fcf011ff6e7140d80/src/mongo/db/modules/atlas/src/disagg_storage/pali/util/shutdown_enforcer.cpp#L41] skips `printAllThreadStacksBlocking()` when `TestingProctor::instance().isEnabled()`. That proctor is enabled by `testingDiagnosticsEnabled=1`, which resmoke passes on **every** run. The comment's stated reason (the dump signals a thread that does not
SERVER-133258Backlog
Add chaos events for log sealing
Claude wrote up some testing gaps in our current chaos events, and the ideas sound good to me:
- *{*}C1 — Standby partition spanning a seal, with apply lag.{*}* Extend the
`seal_during_standby_logd_partition` shape: inject apply-side slowness (or SIGSTOP mongod briefly)before the partition, so when it heals the node's read position is in the sealed segment. Today's chaos heals partitions fast enough that the position is still in the active segment.
- *{*}C2 — Kill/restart between logd-ack a
SERVER-139304ClosedFixed
Pin eviction_updates_trigger=0 in txns_cache_errors tests ahead of WiredTiger default change
WT-17698 changes the WiredTiger default of eviction_updates_trigger from 0 (automatic, half of eviction_dirty_trigger) to 95. The tests under jstests/noPassthrough/txns_cache_errors rely on the old behaviour (update pressure at about 10% of a 256MB cache pulling application threads into eviction, leading to rollback / TemporarilyUnavailable or the cache pressure aborter), so they fail on the 2026-11-13 WiredTiger drop patch (6af5c818e5e64c0007db5bd2).
Failing tests: temporarily_unavailable_erro
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-135691In Progress
Cutback to LS mechanism wiring
SERVER-135689 introduced the ORP fallback wiring and a naive implementation of the LS wiring. This ticket will cover the actual cutback mechanism (cutback mid window, use the SLS-7522 first_untruncated_lsn_in_domain hint).
Also implement loop protection for if we bounce between ORP and LS > some times, check if a newer checkpoint exists if it does crash and reload from it. If it doesnt, stay and wait and make the error visible wait for ORP SLOs to kick in to recover.
SERVER-139305ClosedFixed
WiredTigerUtilTest.GetStatisticsValueInvalidKey fails due to WT connection stat keys shift
WiredTigerUtilTest.GetStatisticsValueInvalidKey asserts that looking up a connection statistics key `WT_STAT_CONN_SESSION_OPEN` through a table statistics cursor fails with NoSuchKey.
A table cursor only accepts keys from WT_STAT_DSRC_BASE (2000) upward. Connection keys start at 1000, but there are now more than 1000 of them, so the connection range overlaps the table range.
The test only passed because WT_STAT_CONN_SESSION_OPEN was 1994, just below 2000. [WT-18993|https://github.com/wired
Components
WiredTiger
Fixed Versions
9.1.0-rc09.1.0-rc1026
JAVA-2673Ready for Work
AggregateToCollectionThenFindOperation casting issue
AggregateToCollectionThenFindOperation casts an AsyncReadBinding to an AsyncWriteBinding to handle Aggregations with {{$out}} and none inline MapReduces. It is a safe cast currently but the aim is to refactor and remove the code smell.
Components
Async
SERVER-135219Backlog
Resharding time estimate does not consider building indexes
See SERVER-134899 for a description of the problems with the time estimates in general and what could be done about them in the long term.
This ticket is specifically to do something to improve the time estimate during building indexes in the short term, since this phase's reporting is more broken than cloning/applying.
For example, today we assume that building indexes + applying takes the same amount of time as cloning. Perhaps after this ticket we could assume that building indexes take
SERVER-131961ClosedFixed
Block processing vectorizer can generate deep ABT trees
Generation and manipulation of these trees is done using recursive algorithms that can exhaust the stack space and crash the server
Fixed Versions
9.1.0-rc08.3.169.0.79.1.0-rc1026
GODRIVER-4270Needs Triage
mongo-go-driver - PR #2725: Bump go.opentelemetry.io/otel/sdk from 1.45.0 to 1.47.0
[dependabot[bot]|https://github.com/apps/dependabot] has created [PR #2725: Bump go.opentelemetry.io/otel/sdk from 1.45.0 to 1.47.0|https://github.com/mongodb/mongo-go-driver/pull/2725] in [mongo-go-driver|https://github.com/mongodb/mongo-go-driver]
GODRIVER-4269Needs Triage
mongo-go-driver - PR #2724: Bump go.opentelemetry.io/otel/trace from 1.45.0 to 1.47.0
[dependabot[bot]|https://github.com/apps/dependabot] has created [PR #2724: Bump go.opentelemetry.io/otel/trace from 1.45.0 to 1.47.0|https://github.com/mongodb/mongo-go-driver/pull/2724] in [mongo-go-driver|https://github.com/mongodb/mongo-go-driver]
SERVER-139274In Progress
Pace incremental spilling with byte-count debt
This replaces the at-crossing burst with pacing: each added key accrues its serialized size as debt, and when the debt covers one batch's bytes ({{Spiller::spillBatchBytes()}}) the sorter calls {{spillIncremental()}} once, repaying the debt with the consumed elements' own sizes.
Debt and repayment are in the same units, so the sorted buffer finishes its spill as the filling buffer refills, whatever the keys' sizes; a document held past the check by {{SorterBatchGuard}} leaves a tail bounded by
SERVER-139272In Progress
Sort the filling buffer and begin its spill when it exceeds half the budget
When the filling buffer's usage exceeds half of {{maxMemoryUsageBytes}}, {{NoLimitSorter}} sorts it and calls {{Spiller::beginSpill()}} before swapping buffers, so the sorted buffer's keys are written while the other buffer receives keys; {{beginSpill}} fires {{SpillCallbacks::onSpillBegin}}. The Sorter base's single pool member becomes an array so the second buffer's pool can be engaged at the first incremental spill, and {{supportsIncrementalSpilling()}} is constant per spiller type, so file-b
SERVER-133600ClosedFixed
Users with write permissions to config database can write to any database through MultiUpdateCoordinatorDocument
Fixed Versions
8.3.169.0.78.0.37
GODRIVER-4268Needs Triage
mongo-go-driver - PR #2723: Bump go.opentelemetry.io/otel/trace from 1.45.0 to 1.47.0 in /internal/test/prose
[dependabot[bot]|https://github.com/apps/dependabot] has created [PR #2723: Bump go.opentelemetry.io/otel/trace from 1.45.0 to 1.47.0 in /internal/test/prose|https://github.com/mongodb/mongo-go-driver/pull/2723] in [mongo-go-driver|https://github.com/mongodb/mongo-go-driver]
GODRIVER-4267Needs Triage
mongo-go-driver - PR #2722: Bump go.opentelemetry.io/otel/trace from 1.45.0 to 1.47.0 in /internal/test/compilecheck
[dependabot[bot]|https://github.com/apps/dependabot] has created [PR #2722: Bump go.opentelemetry.io/otel/trace from 1.45.0 to 1.47.0 in /internal/test/compilecheck|https://github.com/mongodb/mongo-go-driver/pull/2722] in [mongo-go-driver|https://github.com/mongodb/mongo-go-driver]