roachtest: tpchvec/perf failed #82776
Assignees
Labels
branch-release-22.1
Used to mark GA and release blockers, technical advisories, and bugs for 22.1
C-test-failure
Broken test (automatically or manually discovered).
O-roachtest
O-robot
Originated from a bot.
release-blocker
Indicates a release-blocker. Use with branch-release-2x.x label to denote which branch is blocked.
T-sql-queries
SQL Queries Team
Comments
cockroach-teamcity
added
branch-release-22.1
|
Pretty sure #82651 is to blame. |
craig bot
pushed a commit
that referenced
this issue
Jun 13, 2022
82548: randgen: fix postgres mutator for session vars r=otan a=rafiss fixes #67791 Release note: None 82693: rowexec: improve join reader memory usage when ordering is maintained r=yuzefovich a=yuzefovich This commit improves the join reader behavior when used by lookup joins when ordering is maintained. In that case, we have a disk-backed row container that we can spill to disk in order to free up some memory. Previously, we would do so only in the case when the memory error occurs inside of the strategy but would make the query error out if the memory reservation is denied in the span generator or in the join reader itself. This commit makes it so that in all places where the memory accounting is performed, we try to ask the strategy to handle the memory reservations. This allows the join reader ordering strategy attempt spilling looked up rows to disk in all scenarios first, before erroring out the query. Additionally, this commit makes it so that the memory monitor used by that disk-backed row container doesn't hold on to any memory when the corresponding memory account is cleared. By default, our memory monitor can keep a margin of 100KiB in reserve, but this behavior doesn't make sense when we're forcing the disk-backed container to use disk. Release note: None 82749: schemachanger: eager statement phase r=ajwerner a=postamar This commit reworks how the declarative schema changer performs DROPs by removing support for synthetic descriptors in the statement transaction. In-transaction operations are now executed eagerly in the statement phase instead of being deferred to the pre-commit phase. The most significant change this entails is to the rules definitions, since we now no longer have the TXN_DROPPED status. I have taken this opportunity to rewrite them in a way that involves more element set definitions, whose properties are enforced by init-time assertions. The operation implementations in scexec have been simplified somewhat, not having to concern ourselves with synthetic descriptors any more. The opgen definitions are also simplified now that they are no longer peppered with pre-commit phase constraints. Instead, phase constraints are implied by the number of transitions: more than 1 and those other than the first are implied to be restricted to post-commit. This makes all phase constraints implicit. This commit improves the graph validation logic to detect all cycles, not just dep-edge cycles, and renders any detected cycle in the detail of the resulting error. This commit makes some changes in the test dependencies which mock the catalog API, as this now has to support both the in-txn and committed state of the catalog. In the process this commit adds better support for descriptor modification times. Release note: None 82751: scripts: adjust bump-pebble.sh to add build/bazelutil/distdir_files.bzl r=nicktrav a=jbowens The bump-pebble.sh script should `git add build/bazelutil/distdir_files.bzl` which requires updates whenever the Pebble version is bumped. Release note: None 82757: kvserver: implied GCThreshold should be computed from readAt r=ajwerner a=adityamaru `checkGCThresholdAndLeaseRLocked` checks a batch request against a replicas' lease and GCThreshold. When computing the implied GCThreshold we were incorrectly using the clock timestamp at which the lease for the request was evaluated, instead of the timestamp at which the cached protected timestamp state was last updated. The cached pts state is updated everytime the replica observes a fresher protected timestamp state when being considered for GC. In order to uphold the invariants provided by the protected timestamp subsystem, GC decisions must be made based on the timestamp at which the PTS state was last `readAt`. This is akin to how the GC queue picks its implied GCTimestamp in `checkProtectedTimestampsForGC`. Concretely, this bug manifested as an ExportRequest failing with a `BatchTimestampBeforeGCError` even though a protected timestamp record had been written and reconciled. This was a consequence of the implied GCThreshold not taking into account the PTS state. Informs: https://github.com/cockroachlabs/support/issues/1638 Release note: None 82829: distsql: create LeafTxn eagerly for local flow of a distributed plan r=yuzefovich a=yuzefovich This commit makes it so that we eagerly create LeafTxns for local flows of distributed plans. I believe it was an oversight in d6077d5 (added during 21.2 release cycle) where we started creating LeafTxns eagerly when the local flow has concurrency. We can do the same thing for local flows of distributed queries since we know that they must use LeafTxns anyway. This oversight was recently exposed by 6c88496 where some processors began storing the txn internally, rather than accessing it from the flow context. The thing is that we do update the references to the correct txn from the flow context, but it is done _after_ the flow setup has been complete, at which point some processors might have already captured the wrong txn. In particular, this is the case for a local flow of a distributed query with index / lookup joins when the streamer API is disabled. Fixes: #82775. Fixes: #82776. Fixes: #82777. Fixes: #82783. Release note: None Co-authored-by: Rafi Shamim <[email protected]> Co-authored-by: Yahor Yuzefovich <[email protected]> Co-authored-by: Marius Posta <[email protected]> Co-authored-by: Jackson Owens <[email protected]> Co-authored-by: Aditya Maru <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
roachtest.tpchvec/perf failed with artifacts on release-22.1 @ 3181b7faa2f3b41d6a15ab4b74d2c60bcfe5132d:
Help
See: roachtest README
See: How To Investigate (internal)
This test on roachdash | Improve this report!
Jira issue: CRDB-16651
The text was updated successfully, but these errors were encountered: