Re: Deadlock? idle in transaction

Поиск
Список
Период
Сортировка
От Stephan Szabo
Тема Re: Deadlock? idle in transaction
Дата
Msg-id Pine.BSF.4.21.0110111303320.93237-100000@megazone23.bigpanda.com
обсуждение исходный текст
Ответ на Deadlock? idle in transaction  (Michael Meskes <meskes@postgresql.org>)
Ответы Re: Deadlock? idle in transaction  (Michael Meskes <meskes@postgresql.org>)
Список pgsql-hackers
On Thu, 11 Oct 2001, Michael Meskes wrote:

> A customer's machine hangs from time to time. All we could find so far is
> that postgres seems to be in state "idle in transaction":
> 
> postgres 19317  0.0  0.3  8168  392 ?        S    Oct05   0:00 /usr/lib/postgresql/bin/postmaster -D
/var/lib/postgres/data
> postgres 19983  0.0  0.8  8932 1020 ?        S    Oct05   0:01 postgres: postgres rabatt 192.168.50.222 idle in
transaction
> postgres 21005  0.0  0.0  3484    4 ?        S    Oct06   0:00 /usr/lib/postgresql/bin/psql -t -q -d template1
> postgres 21014  0.0  0.7  8892  952 ?        S    Oct06   0:01 postgres: postgres rabatt [local] VACUUM waiting
> postgres 21833  0.0  0.4  3844  572 ?        S    Oct06   0:00 /usr/lib/postgresql/bin/pg_dump rabatt
> postgres 21841  0.0  1.2  9716 1564 ?        S    Oct06   0:00 postgres: postgres rabatt [local] COPY waiting
> postgres 22135  0.0  0.9  8856 1224 ?        S    Oct06   0:00 postgres: postgres rabatt 192.168.50.223 idle in
transactionwaiting
 
> 
> I'm not sure what's happening here and I have no remote access to the
> machine myself. Any idea what could be the reason for this?
> 
> There may be some client processes running at the time the dump and the
> vacuum commands are issued that have an open transaction doing nothing. That
> is the just issued a BEGIN command. Thinking about it run some inserts at
> the very same time, although that's not likely.
> 
> Any hints are appreciated. Thanks in advance.

Well, it'd be likely to get in this state if the first transaction grabbed
any write locks and then sat on them without committing or doing any more
commands, since the vacuum would wait on that and the rest of the
transactions will probably wait on the vacuum.  Is that a possible
situation?





В списке pgsql-hackers по дате отправления:

Предыдущее
От: Bruce Momjian
Дата:
Сообщение: Re: Unicode combining characters
Следующее
От: Bruce Momjian
Дата:
Сообщение: Re: CLUSTER TODO item