-
Notifications
You must be signed in to change notification settings - Fork 1.1k
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
[docdb] QLStressTest.OldLeaderCatchUpAfterNetworkPartition flaky in master #3465
Comments
passing this to you @robertsami since you were looking at another QLStressTest recently. edit: actually, looking at the 100 buckets view, seems like stability for this has improved recently: https://detective-gcp.dev.yugabyte.com/stability/test?buckets=100&class=QLStressTest&name=OldLeaderCatchUpAfterNetworkPartition not sure if this is still a concern |
Can confirm this is still a problem, seems like 11/25 commit failure rate. |
Summary: Prevoiusly this test would fail with two kinds of transient failures: 1) Assertion that chosen leader had an election term of 1 2) Assertion that disconnected leader's election term did not advance To address 1), we relax this assertion, since it should be expected that the chosen leader's election term could be greater than 1 by the time we get to choosing it. This does not affect the validity of the test. 2) was caused by unbroken incoming connections to the "isolated" leader. When we set a tserver to isolated, we call TEST_Isolated, which had a bug whereby incoming connections were not properly terminated. This diff corrects that behavior. Test Plan: `ybd --cxx-test client_ql-stress-test --gtest_filter QLStressTest.OldLeaderCatchUpAfterNetworkPartition --clang -n 100` Reviewers: sergei Reviewed By: sergei Subscribers: ybase, bogdan Differential Revision: https://phabricator.dev.yugabyte.com/D9628
last 15 commits seem good for this test after 1274b42 -- closing for now |
https://detective-gcp.dev.yugabyte.com/job/github-yugabyte-db-phabricator/23596/artifact/build/debug-clang-dynamic-ninja/yb-test-logs/tests-client__ql-stress-test/QLStressTest_OldLeaderCatchUpAfterNetworkPartition.log
Expected: (pre_isolate_op_id.index) > (key), actual: 1050 vs 1050
Expected: (leader) != (nullptr), actual: NULL vs 8-byte object <00-00 00-00 00-00 00-00>
The text was updated successfully, but these errors were encountered: