Re: How to diagnose application "hangs" in pg_receive?
От | Clemens Eisserer |
---|---|
Тема | Re: How to diagnose application "hangs" in pg_receive? |
Дата | |
Msg-id | BANLkTi=LLpnGZOsyHQFnYvpAe1C7T8+HUQ@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: How to diagnose application "hangs" in pg_receive? ("Kevin Grittner" <Kevin.Grittner@wicourts.gov>) |
Список | pgsql-jdbc |
Hi Kevin, If you feel personally attacked, just because I highlighted that I am sure the problem is not caused by a query taking a long time to execute, I am sorry. Nontheless, please find my detailed answers below. Thanks, Clemens > You haven't told us what version of PostgreSQL you're running, PostgreSQL 8.4.7 postgresql-8.4-701.jdbc3.jar > spite of my asking that you post the results of a query which would > have told us that and other useful information. Because as I said, the problem is not the execution of the query. Last time I experienced such a hang it was a simple "select .... from ... where id=x" query, which executes usually in a few ms. > The only evidence that you have provided suggests that at > least part of your problem has been bloat, but you dismissed that > out of hand without showing any sign that you understand that issue > or have checked for it. The database has about 2MB of data stored and is mostly read-only. I know what bloat is, understand the issue and have given a clear indication that I know bloat is not the problem. Therefor narrowing down the possibilities, and to direct guessing into another direction. > In follow-ups, please try to avoid hand-wavy terms like "forever"; > tell us whether you waited a minute, an hour, ten days, whatever > "forever" actually was, along with any observations on CPU and disk > usage during this time. You didn't show the active query in your > listings -- were those taking during the time the query was hanging? Forever means, what forever usually means - as long as I terminate the client. I took this hand-wavy term to make clear that my understanding of the problem is that its not a slowly running query.
В списке pgsql-jdbc по дате отправления: