Brandon Nesterenko <brandon.nesterenko@mariadb.com> writes:
I see your line of thinking; agreed that --skip-slave-errors=all with optimistic parallel replication remains problematic. Though I think we should open it as a different JIRA, as the use cases differ, i.e.
Ok, that's fine. Right, I saw one of Roel's tests in MDEV-27512 was parallel replication and --slave-skip-errors; but I agree that this is different from your patch for the non-parallel replication case.
likely wouldn't be able to see the fix for the parallel replica through in time for this release, as I'm on vacation April 21 - May 3). Though I'm happy to still take on this new issue.
That shouldn't be a problem, this bug is something I introduced all the way back to when optimistic parallel replication was implemented, I didn't consider --skip-slave-errors at all :-(. Surely it can wait a few weeks more. But it's great if you want to look into it when you're back, much appreciated. - Kristian.