| aaron added a comment. |
In T200420#4453134, @Addshore wrote:Something to note, because the locks are no longer in the DB, we end up selecting the same 15 or so wikis that are locked all of the time.
It could be that the other wikis actually don't have locks:before using the redis lock manager the status of the lock from the db was also in the select so that locked dbs would not be selected at all.
You mean the condition on chd_touched in getCandidateClients()? That would roughly correlate with lock state.
It's odd that engageClientLock() is blocking for the SQL coordinator but not the LockManager-using subclass (Database::lock is 5s by default). That would make the LM-based selectClient() more likely to not find anything and return false (if there are a few busy wikis), as it only does one pass with no usleep(). Since dispatchMaxTime is not PHP_INT_MAX (but 720) and I don't see --max-passes in puppet, then max-passes for dispatchChanges.php runs is 1, right? Thus, once selectClient() returns false, then dispatchChanges.php would exit afaik. The loop/delay (idle-delay) logic would not happen in such cases.
Cc: zeljkofilipin, aaron, gerritbot, Stashbot, WMDE-leszek, Addshore, TerraCodes, Liuxinyu970226, matej_suchanek, Aklapper, Gaboe420, Versusxo, Majesticalreaper22, Giuliamocci, Adrian1985, Cpaulf30, Lahi, Gq86, Baloch007, Darkminds3113, Bsandipan, Lordiis, GoranSMilovanovic, Adik2382, Th3d3v1ls, Ramalepe, Liugev6, QZanden, LawExplorer, Lewizho99, Maathavan, Jonas, Wikidata-bugs, aude, Lydia_Pintscher, Mbch331
_______________________________________________ Wikidata-bugs mailing list Wikidata-bugs@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs
