Обсуждение: Slow UPDATE in logs that's usually fast
Hello,
There's a slow UPDATE query in my logs (~60s). However, when I run it manually, it's really fast ( < 0.5ms).
2016-11-23 18:13:51.962 GMT [742-25]: bpd_production bpd@web001(40916) 00000 Passenger RubyApp: /var/www/bpd/current (production) LOG: duration: 59876.947 ms statement: UPDATE "contacts" SET "updated_at" = '2016-11-23 18:12:52.055456' WHERE "contacts"."id" = 2845179
This particular query isn't waiting on any locks that I can see. I actually found it because I followed a lock queue up the chain and it bubbled up to this query. This was one that was waiting on process 742:
2016-11-23 18:13:00.095 GMT [314-118]: bpd_production bpd@web001(40547) 00000 Passenger RubyApp: /var/www/bpd/current (production) LOG: process 314 still waiting for ShareLock on transaction 1663649998 after 1000.067 ms
2016-11-23 18:13:00.095 GMT [314-119]: bpd_production bpd@web001(40547) 00000 Passenger RubyApp: /var/www/bpd/current (production) DETAIL: Process holding the lock: 742. Wait queue: 314.
2016-11-23 18:13:00.095 GMT [314-120]: bpd_production bpd@web001(40547) 00000 Passenger RubyApp: /var/www/bpd/current (production) CONTEXT: while updating tuple (288387,8) in relation "contacts"
2016-11-23 18:13:00.095 GMT [314-121]: bpd_production bpd@web001(40547) 00000 Passenger RubyApp: /var/www/bpd/current (production) STATEMENT: UPDATE "contacts" SET "news_items_last_modified" = '2016-11-23 18:12:59.090806' WHERE "contacts"."id" = 2845179
Anyhow, I'm at a complete loss here since I've hit a dead end.
Thank you!
Aldo Sarmiento