-
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
*: disable merge queue in more tests #29235
Conversation
If this LGTY, please send this through Bors to reduce the flakiness! |
pkg/internal/client/db.go
Outdated
@@ -471,6 +471,7 @@ func (db *DB) AdminMerge(ctx context.Context, key interface{}) error { | |||
// | |||
// The keys can be either byte slices or a strings. | |||
func (db *DB) AdminSplit(ctx context.Context, spanKey, splitKey interface{}) error { | |||
return errors.New("manual split") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
left over?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hahaha, thanks. That's going to cause a test failure or fifty.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
but please circle back to my comment about the skipped test.
bors r+
Reviewed 19 of 21 files at r1, 2 of 2 files at r2.
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale)
pkg/sql/distsql_physical_planner_test.go, line 99 at r2 (raw file):
func TestPlanningDuringSplitsAndMerges(t *testing.T) { defer leaktest.AfterTest(t)() t.Skip()
Add a comment. Or unskip this? Might be spurious.
Build failed |
bors r+ TestLint failed with the url checker :( |
I didn't look at the contents of this PR (yet), but be aware that there is also |
Build failed (retrying...) |
The current test failure looks like it's still caused by the merge queue: https://teamcity.cockroachdb.com/viewLog.html?buildId=869108&buildTypeId=Cockroach_UnitTests
I'll merge the other PR. Let's smoke out the remaining flakes without messing up everyone else's day. |
Build failed |
I did, yep! TestImportPgDump is finding a real bug with merges, I think. I've plumbed in a DisableMergeQueue knob with a TODO for now. |
29230: server: deflake TestRapidRestarts r=benesch a=petermattis In go1.10 and earlier, it was not safe to call `http.ServeMux.ServeHTTP` concurrently with `http.ServeMux.Handle`. (This is fixed in go1.11). In the interim, provide our own safeServeMux wrapper that provides proper locking. Fixes #29227 Release note: None 29236: Revert "storage: enable the merge queue by default" r=tschottdorf a=benesch This reverts commit 98ca1d0. The merge queue will be reenabled once flaky tests are fixed. To reviewers: I'd much prefer to merge #29235. But if that gets stuck in code review or the flakiness reaches a breaking point, feel free to merge this instead. Co-authored-by: Peter Mattis <[email protected]> Co-authored-by: Nikhil Benesch <[email protected]>
bors r=tschottdorf |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale)
pkg/sql/distsql_physical_planner_test.go, line 99 at r2 (raw file):
Previously, tschottdorf (Tobias Schottdorf) wrote…
Add a comment. Or unskip this? Might be spurious.
Oops, didn't mean to commit. It's removed in the rev that's getting bors'd.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Did you look at tests using kv.setupMultipleRanges()
? E.g. TestTxnCoordSenderRetries
should be in here but it's not.
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Also, did you check if any flake issues can be closed? "No" is a good enough answer, but I'd like to know for when looking at the ones assigned to me.
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale)
Agh, I thought those were all covered by
|
There was only the one I linked in the PR description when I opened this last night. I'll do another sweep sometime this afternoon. |
Take a more aggressive approach to disabling the merge queue in tests. This commit disables the merge queue in all tests that send a manual AdminSplit request, unless those tests were explicitly verified to be safe to run with the merge queue on. Release note: None
This is contingent on #29324 going in first. |
bors r=tschottdorf |
1 similar comment
bors r=tschottdorf |
29235: *: disable merge queue in more tests r=tschottdorf a=benesch Take a more aggressive approach to disabling the merge queue in tests. This commit disables the merge queue in all tests that send a manual AdminSplit request, unless those tests were explicitly verified to be safe to run with the merge queue on. Release note: None Fix #29085. Fix #29242. Co-authored-by: Nikhil Benesch <[email protected]>
Build succeeded |
Take a more aggressive approach to disabling the merge queue in tests.
This commit disables the merge queue in all tests that send a manual
AdminSplit request, unless those tests were explicitly verified to be
safe to run with the merge queue on.
Release note: None
Fix #29085.
Fix #29242.