https://bugs.kde.org/show_bug.cgi?id=522845
--- Comment #3 from Michal Schorm <[email protected]> --- (In reply to Allen Winter from comment #2) > Michal, > > Would you be so kind and forward my patch to the mariadb experts for review? > https://invent.kde.org/pim/akonadi/-/merge_requests/400 > > I do see that `skip_slave_start` has been deprecated in favor of > `skip_replica_start` in some mysql version. I tried `skip_replica_start` > here and it didn't work. so I wonder what to do about that. MariaDB currently only has the old name `skip_slave_start`. The adoption of the more inclusive language is tracked as https://jira.mariadb.org/browse/MDEV-18777 but currently stalled. In MySQL 9.7 the old name AFAIK still works, though deprecated. --- I thought about whether the issue I'm reporting here is Fedora specific. Turns out it is an universal problem across Linux distributions for a long time - either Akonadi crashes on short MariaDB / MySQL timeout, or needs to deal with various post-DB-crash issues after the timeout kills on subsequent starts. --- I believe the merge request you prepared is very reasonable. The `skip_slave_start` was already acked on MariaDB upstream in the discussion prior my report here. Increasing the timeout to 30s looks reasonable to be and should help universally across distros. Although I see distros converting to SQLite backend (Gentoo, Arch, Ubuntu). Do I get it right that SQLite backend is now the preferred backed by Akonadi upstream? -- You are receiving this mail because: You are watching all bug changes.
