Known Issues
5279 tracked MongoDB server issues from Jira with status, affected versions, and resolution details.
SERVER-139470Open
Add initial alerts for replicated size and count dashboards
Figure out the infrastructure to set up alerts for related grafana dashboards we built. Start with oplogLagSecs and tailer/flusher health.
Start with slack channel notification, with capped alert frequency.
SERVER-139281In Code Review
Suppress static Windows tcmalloc import warnings
Static Windows links of {{tcmalloc_minimal}} produce LNK4217/LNK4286 because the public headers declare DLL imports while the allocator comes from a static archive. Propagate {{/IGNORE:4217}} and {{/IGNORE:4286}} through this target's {{linkopts}} only for Windows with static linking. Keep the allocator declarations unchanged so the existing executable exports remain available to consumers.
Confined to {{src/third_party/gperftools/BUILD.bazel}}. Prerequisite for #66298, independently based on m
Components
Bazel
Fixed Versions
9.1.0-rc09.1.0-rc1026
SERVER-85195ClosedFixed
enable launching MongoT for sharded cluster deployment
This ticket depends on SERVER-85194 because that ticket will add the MongoTFixture and MongotLauncher classes.
As with SERVER-85194, this ticket will require having a mongot binary locally downloaded to your VM. This ticket will create the new sharded_search_end_to_end suite and a dummy jstest that launches a sharded cluster with a mongot for each mongod.
The constructor for [ShardedClusterFixture|https://github.com/10gen/mongo/blob/7fad9f533469ac953dd8d18b4c44c45398121aa6/builds
Fixed Versions
8.0.0-rc0
SERVER-85198ClosedFixed
create a MongoTFixture for resmoke and enable launching mongot in a single node repl set
In ReplSetBuilder, the launch_mongot flag will be checked. The value of that flag will be [passed in _new_mongod()|https://github.com/10gen/mongo/blob/42ecf946801bcf31c6d20f956c73bce607a5f607/buildscripts/resmokelib/testing/fixtures/_builder.py#L201], which creates a fixture container for mongod (with specified configuration/parameters) fixture(s) in it. That value will eventually be saved to [mongod_options|https://github.com/10gen/mongo/blob/master/buildscripts/resmokelib/testing/fixtures/_bui
Fixed Versions
8.0.0-rc0
SERVER-139247Blocked
Investigate impact of disagg on WT assumptions in the JOO cost model
Investigate which WiredTiger assumptions baked into the JOO cost model ({{{}join_cost_estimator_impl.cpp{}}}, {{{}join_estimates.cpp{}}}, {{{}catalog_stats.cpp{}}}, {{createCatalogStats()}} in {{{}executor.cpp{}}}) break under disaggregated storage and investigate if this can lead to poor plan quality.
*Assumptions that could be wrong*
* {_}Random IO is 5.4x sequential IO{_}{*}.{*} Disagg has no sequential locality.
* {_}Per-page IO costs are ~160us sequential and ~864us random{_}{*}.{*}
SERVER-138772Backlog
Produce delivery plan and architecture update from conflict detection prototype findings
h2. Overview
Consolidate the prototype findings into a high-confidence delivery plan and technical architecture for decoupling conflict detection from the storage engine.
h2. Scope of Work
* Update the architecture document with findings from the prototype tickets in this epic.
* Finalize the prototype API headers (conflict detector, staged writes / {{StagedWriteOverlayCursor}}, primary oplog application) as the starting point for follow-on production work.
* Ticket breakdowns and estimates for
SERVER-138764Backlog
Validate read-your-writes across RecordStore, SortedDataInterface and Container with staged writes
h2. Overview
Validate that RecordStore, SortedDataInterface and container behavior is unchanged when backed by staged writes and the {{StagedWriteOverlayCursor}}, and that transactions read their own writes through collection and index scans.
h2. Scope of Work
* Run the existing RecordStore, SortedDataInterface and container test harnesses in staged mode.
* jstests for read-your-writes via collection scans, index scans, point lookups and {{updateWithDamages}}.
* Multi-update/multi-delete and {{
SERVER-138763In Progress
Prototype StagedWriteOverlayCursor for staged writes
h2. Overview
Prototype the {{StagedWriteOverlayCursor}}, which presents a WiredTiger table as the union of its committed contents (at the transaction's snapshot) and the RecoveryUnit's staged entries for that table. The cursor is opaque to key and value contents; RecordStore, SortedDataInterface and container code above it is unchanged.
h2. Scope of Work
* Semantics per design note section 6.1: exact search, next/prev that overlays staged entries on the table in key order, with staged entries w
SERVER-139308Backlog
Complete TODO listed in SERVER-138483
There is a TODO in code owned by your team which references a resolved ticket.
The TODO appears in:
* src/mongo/db/query/stage_builder/sbe/gen_group.cpp:1742
You can also search for all references to the resolved ticket here:
https://github.com/10gen/mongo/search?q=SERVER-138483&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 th
SERVER-137762Blocked
Re-evaluate if we should run join workloads with plan cache off
These variants currently run with join plan cache disabled:
* Query Join Optimization 3-Node ReplSet ARM (Release Policies, No Shedding)
* Query Join Optimization 3-Node ReplSet ARM (Persistent Stats 1k, Release Policies, No Shedding)
* Query Join Optimization 3-Node ReplSet ARM (Persistent Stats 2k, Release Policies, No Shedding)
Should we run tests in these variants with plan cache on instead?
SERVER-139361Backlog
Charge array filter merge cache memory to the operation memory tracker
h2. Problem
UpdateArrayNode keeps every merged update subtree in {{_mergedChildrenCache}} for the lifetime of the update tree. In a {{multi: true}} update the cache persists across documents. SERVER-138477 bounds the work per document and clears all caches at 1000 entries, but the clear threshold counts entries, not bytes. An entry is a deep copy of a merged tree, so its size grows with the number of array filters in the merge chain, and nothing caps that number. Memory held between documents i
SERVER-139003Open
Report specific join optimization fallback reasons from singleTableAccessPlans()
Every failure in {{singleTableAccessPlans()}} is reported as {{failedToGetSingleTableAccessViaCBR}} by the executor. That includes:
* planning or CBR errors
* CBR accepting several plans it can't rank
* the single plan CBR can't estimate (SERVER-138000)
SERVER-134637 will add another: falling back when a node's chosen plan uses a hashed, wildcard or collated index. These cases look the same in serverStatus and {{{}$queryStats{}}}.
Consider having {{singleTableAccessPlans()}} return a
SERVER-139099ClosedDuplicate
Add a fallback metric for when JOO bails due to a hashed/wildcard/collated index
SERVER-134637 added support for costing hashed/wildcard/collated indexes using CBR but explicitly disabled these for JOO such that JOO bails out if the winning plan for the base collection has one of these index types.
This is to add a fallback metric that provides observability when this happens.
SERVER-138160Open
Investigate cost model changes needed to pick cheaper INLJ on small table
For TPCE locust queries q08/q10/q14/q15/q17 it seems we prioritize a plan with a HJ/NLJ on a COLLSCAN instead of an INLJ probing on the "smaller" table.
When comparing to classic, the classic plan is (closer) to an INLJ (probing on the small table). While the absolute difference is small (on the order of ~10s of docs) this cost is visible in terms of query latency because each query is very fast.
SERVER-121078Open
Join optimization: non-join-optimization-enabled NLJ execution may win against HJ in the case of $limit
In the case of a small $limit, the non-join-optimized execution can terminate the processing of the entire query immediately once the $limit is reached. However, if join optimization elects to build any hash tables, those tables need to be built in their entirety, incurring a fixed cost regardless of the $limit.
To reproduce, run the following query against the tpch 0.1 dataset:
{code:java}
{">>>command": {"aggregate":"lineitem","pipeline":[{"$lookup":{"from":"orders","localField":"l_order
SERVER-139071ClosedGone away
Collection metadata recoverer incorrectly marks untracked collections on the config database as "unowned"
The DB primary of the config database is the config server. When recovering authoritative metadata, the config server should mark each untracked collection in the config database as "untracked", but instead it [marks it as unowned|https://github.com/10gen/mongo/blob/d66cb87b8981b27cfa97c374d02c24cb2f5ac4b6/src/mongo/db/shard_role/shard_catalog/collection_metadata_recoverer.cpp#L358-L376]. This doesn't have much consequence beyond [making CMC complain about it|https://github.com/10gen/mongo/blob/
SERVER-119957Open
Investigate whether SBE plans generated by the join optimizer yield frequently enough
The join optimizer produces new kinds of SBE plans which may have different data access patterns than plans generated in the past. We should investigate whether the SBE engine's yielding behavior is sufficient, or whether it is possible for the join optimizer to generate a plan that (depending on the shape of the data) may not yield for seconds or minutes. My recollection is that SBE plans currently trigger yielding at the leave stages doing data access ({{scan}} or {{ixscan}}), but perhaps it i
SERVER-133099ClosedFixed
Add server parameter and allow to drop system.session for a sharded cluster
Add a new undocumented parameter to enable the drop of `system.session` collection on a sharded cluster.
The parameter is only meant to be used by specialized operators by following a documented playbook.
To serialize with potential concurrent creations (the collection is forced re-created every 5m) we guarantee to re-use the same shard as coordinator (i.e the first shard alphabetically).
Considering `addShard` or `removeShard` could potentially change the first shard alphabetically, the play
Components
Upgrade/Downgrade
Fixed Versions
9.1.0-rc09.1.0-rc1026