Follow-up from the daily CI check (machine-written). This test was today's stabilisation candidate. It is red on stable-18.4.x, stable-17.10.x and stable-18.8.x, 88 failures in 570 executions over 28 days, and the most recent failures are all in stable-18.4.x / stable-17.10.x / stable-18.8.x on chrome / mysql latest / tomcat 11-jdk25. F2 fails 22.7% of runs in that configuration on stable-18.4.x.
Local measurement, before any change (stable-18.4.x, 18.4.6-SNAPSHOT, chrome, hsqldb_embedded, jetty_standalone, one provisioned wiki reused across runs):
executions
passed
failed
pass rate
20
20
0
100%
20 clean executions bound the failure rate below 15% (95%, rule of three). That is compatible with the rate CI sees outside the concentrating configuration, so the flicker did not reproduce here. Without a reproduction, no before/after comparison could show that a change helps, so no fix PR was opened. The integration tests were not run in CI; this measured one test on one machine.
What a next attempt needs is the concentrating configuration: tomcat 11 on jdk25 with MySQL (a fresh build per repetition), where 20 executions at ~20% would be expected to fail about 4 times. The fix idea from the previous comment (waitUntilLocalChangesArePushed() before each tab switch) is still unmeasured.
This message was sent by Atlassian Jira (v9.3.0#930000-sha1:287aeb6)
If image attachments aren't displayed, see this article.
