Each line is a new regression – i.e. a new issue not previously seen. Previously seen issues (i.e. open defects already logged) were/are removed by our filters. Also, this list lists a maximum of 10 trials per issue, even though there may have been more.

Technical sidenode; All this is done by pquery-go-expert.sh which is an advanced way of running the pquery framework. It 1) preps testcase reducers (reducer.sh) ready-to-run for each trial, 2) removes >10 ‘issue duplicates’ leaving 10 trials per issue (10 trials is usually more then sufficient to get a good test case), 3) filters known issues (using pquery-clean-known.sh which in turns uses known_bugs.strings – our bug filter list). If you ever have a crash, run text_string.sh against it and compare it with the list in known_bugs.strings (even old bugs are left, but remarked when fixed) – it’s a great way to find out if a bug is already been logged for the crash you’re seeing. For all these scripts, just clone percona-qa (https://github.com/Percona-QA/percona-qa) from Github.

In other words/in summary, the list above is substantial proof of a lot of new regressions in 8.0. Still, I have trust in Oracle to resolve most of these before GA!