| Bug #89508 | MTR test rpl.rpl_public_key is unstable | ||
|---|---|---|---|
| Submitted: | 2 Feb 2018 8:35 | Modified: | 24 Apr 2018 9:53 |
| Reporter: | Laurynas Biveinis (OCA) | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | Tests: Replication | Severity: | S3 (Non-critical) |
| Version: | 8.0.4 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
| Tags: | mtr | ||
[2 Feb 2018 14:43]
Erlend Dahl
Thanks for the feedback, we have observed the same problem internally.
[23 Apr 2018 11:30]
Erlend Dahl
Bug#90558 rpl.rpl_public_key test fails was marked as a duplicate.
[24 Apr 2018 9:52]
Margaret Fisher
Posted by developer: Test case, so no changelog entry needed, but thanks to Laurynas Biveinis for the analysis and fix suggestion.
Description: Under load, occasionally: rpl.rpl_public_key 'row' w2 [ fail ] Test ended at 2018-01-29 05:35:26 CURRENT_TEST: rpl.rpl_public_key --- /mnt/workspace/mysql-5.7-asan-param/BUILD_TYPE/release-asan/Host/xenial-x64-bigram/mysql-test/suite/rpl/r/rpl_public_key.result 2018-01-29 07:26:08.481358650 +0300 +++ /mnt/workspace/mysql-5.7-asan-param/BUILD_TYPE/release-asan/Host/xenial-x64-bigram/build/mysql-test/var/2/log/rpl_public_key.reject 2018-01-29 08:35:24.742590337 +0300 @@ -65,7 +65,6 @@ SELECT * FROM t1; c1 1 -2 ==== Reset and cleanup ==== [connection slave] include/stop_slave.inc mysqltest: Result content mismatch How to repeat: Not sure, rpl_public_key with high --repeat --parallel? I have sen this also on ASan/UBSan combined build, which is slower still Suggested fix: By testcase analysis the cause appears to be START SLAVE immediately followed by SELECT, without syncing the slave first. Thus the fix would be adding sync_slave_sql_with_master.inc between the two.