From: | Christopher Browne <cbbrowne(at)acm(dot)org> |
---|---|
To: | pgsql-novice(at)postgresql(dot)org |
Subject: | Re: speeding up pg_dump? |
Date: | 2005-12-28 04:51:02 |
Message-ID: | m3k6dpq0y1.fsf@mobile.int.cbbrowne.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-novice |
> From: "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us>
>> Every transaction takes ExclusiveLock on its own transaction ID. That
>> in itself isn't an issue. You sure you don't see any rows with granted
>> = 'f' while pg_dump is running and everything seems blocked?
>
> yes. during a pg_dump, there are like 30 locks - all of then granted (t)
That makes sense; once you are dumping the 30th table, there will be
about 30 locks, although they should only be AccessShared locks.
> i'll set up pg8.1.1 tomorrow on a new server to check if its db/web or
> server related...
You can expect to see a bunch of AccessShared locks associated with
the transaction used for the pg_dump.
The interesting question is what *else* is trying to grab a lock; that
"something else" is presumably the root of your troubles.
--
wm(X,Y):-write(X),write('@'),write(Y). wm('cbbrowne','gmail.com').
http://linuxfinances.info/info/slony.html
Keeping instructions and operands in different memories saves .20
(.09) microseconds.
From | Date | Subject | |
---|---|---|---|
Next Message | Kevin Crenshaw | 2005-12-28 12:15:36 | Re: Complex Query Help- For Me, Anyway |
Previous Message | Yumiko Izumi | 2005-12-28 02:43:05 | We want to monitor total size of database |