https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6444
--- Comment #29 from peter gervai <[email protected]> --- (In reply to Kevin A. McGrail from comment #28) > Recommended patches For me to be able to provide you with a patch would require a complete analysis of the SQL code, and most probably rewriting large chunks of it, including the PGSQL parts. > because to me this seems like a slamdunk for InnoDB or a Redis Backup? It's not a really hard thought excercise to realise and if I wanted to use an inferior system just because there's poor coding in the parts in question then I wouldn't be still here, right? I haven't worked with this code and after a *very* brief and ignorant look I only guess that locking wasn't really thought over, since implicit locking seem to conflict on a busy database. I probably still have to look through since I cannot fiddle with it just by poking transactions or explicit locking here and there but it requires time, for me it requires much more than for those familiar with the code, but eventually I'll reach that point. Giving advices with bad intent doesn't really help. (As a sidenote for me it seems that put_tokens (SQL) is always involved in deadlocks, probably by locking large chunks within a transaction, but I may be wrong.) -- You are receiving this mail because: You are the assignee for the bug.
