-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
kv/kvserver/allocator/allocatorimpl: TestAllocatorFullDisks failed #100033
Labels
branch-release-23.1
Used to mark GA and release blockers, technical advisories, and bugs for 23.1
C-test-failure
Broken test (automatically or manually discovered).
GA-blocker
O-robot
Originated from a bot.
T-kv
KV Team
Milestone
Comments
cockroach-teamcity
added
branch-release-23.1
Used to mark GA and release blockers, technical advisories, and bugs for 23.1
C-test-failure
Broken test (automatically or manually discovered).
O-robot
Originated from a bot.
labels
Mar 30, 2023
This test failed after 20 minutes of stressing on master. I'm going to look further into it now.
|
I think this is caused by incorrectly using the shed threshold instead of the rebalance threshold when calculating the total number of ranges for the test - introduced in #97409.
|
craig bot
pushed a commit
that referenced
this issue
Apr 4, 2023
100189: kvcoord: Restart ranges on a dedicated goroutine. r=miretskiy a=miretskiy Restart ranges on a dedicated goroutine (if needed). Fix logic bug in stuck range handling. Increase verbosity of logging to help debug mux rangefeed issues. Informs #99560 Informs #99640 Informs #99214 Informs #98925 Informs #99092 Informs #99212 Informs #99910 Informs #99560 Release note: None 100525: rpc: Handle closed error r=erikgrinaker a=andrewbaptist We close the listener before closing the connection. This can result in a spurious failure due to the Listener also closing our connection. Epic: none Fixes: #100391 Fixes: #77754 Informs: #80034 Release note: None 100528: sql: fix flaky TestSQLStatsCompactor r=j82w a=j82w The test failure is showing more total wide scans than expected. Change the compact stats job to run once a year to avoid it running at the same time as the test. The interceptor is disabled right after delete reducing the possibility of another operation causing a conflict. Epic: none closes: #99653 Release note: none 100589: allocator: deflake full disk test r=andrewbaptist a=kvoli In #97409 we introduced cluster settings to control the disk fullness threshold for rebalancing towards a store and shedding replicas off of the store. The `TestAllocatorFullDisks` assumes the total number of range bytes is equal or less than the rebalance threshold of the nodes, however the test was updated to use the shed threshold instead. This caused the test to flake occasionally as there was more than the expected amount of total range bytes. This patch changes the ranges per node calculation to use the rebalance threshold again, instead of the shed threshold ``` dev test pkg/kv/kvserver/allocator/allocatorimpl -f TestAllocatorFullDisks -v --stress ... 15714 runs so far, 0 failures, over 39m45s ``` Fixes: #100033 Release note: None 100610: roachtest: set config.Quiet to true r=herkolategan a=srosenberg After refactoring in [1], the default of config.Quiet was set to false since the roachprod CLI option is intended to set it to true. This resulted in an unwanted side-effect, namely roachtests running with the new default. Consequently, test_runner's log ended up with a bunch of (terminal) escape codes due to (status) spinner. This change ensures roachtest explicitly sets config.Quiet to true. [1] #99133 Epic: none Release note: None Co-authored-by: Yevgeniy Miretskiy <[email protected]> Co-authored-by: Andrew Baptist <[email protected]> Co-authored-by: j82w <[email protected]> Co-authored-by: Austen McClernon <[email protected]> Co-authored-by: Stan Rosenberg <[email protected]>
blathers-crl bot
pushed a commit
that referenced
this issue
Apr 4, 2023
In #97409 we introduced cluster settings to control the disk fullness threshold for rebalancing towards a store and shedding replicas off of the store. The `TestAllocatorFullDisks` assumes the total number of range bytes is equal or less than the rebalance threshold of the nodes, however the test was updated to use the shed threshold instead. This caused the test to flake occasionally as there was more than the expected amount of total range bytes. This patch changes the ranges per node calculation to use the rebalance threshold again, instead of the shed threshold Fixes: #100033 Release note: None
Will be closed on #100646 |
Closed on #100646 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
branch-release-23.1
Used to mark GA and release blockers, technical advisories, and bugs for 23.1
C-test-failure
Broken test (automatically or manually discovered).
GA-blocker
O-robot
Originated from a bot.
T-kv
KV Team
kv/kvserver/allocator/allocatorimpl.TestAllocatorFullDisks failed with artifacts on release-23.1 @ aec78f33d45a8376a0ecec885688bae60dbfb85c:
Parameters:
TAGS=bazel,gss
Help
See also: How To Investigate a Go Test Failure (internal)
This test on roachdash | Improve this report!
Jira issue: CRDB-26187
The text was updated successfully, but these errors were encountered: