Problem: pg_ctl does not not stop postgres properly even with immediate option

Поиск
Список
Период
Сортировка
От Jürgen Fuchsberger
Тема Problem: pg_ctl does not not stop postgres properly even with immediate option
Дата
Msg-id 51F90E3B.6030604@uni-graz.at
обсуждение исходный текст
Список pgsql-admin
Hi all,

I am on a Debian wheezy system with postgresql 9.1 installed. I have a
hot standby configuration with the standby server doing the job of
taking backups of the database. For this purpose I stop the database,
start rsync, then start the DB again.
This afternoon I did some tests and found that postgres did not stop as
I expected using
> /etc/init.d/postgresql stop

so I tried
> pg_ctlcluster --force 9.1 main stop

Still without success, next:
> pg_ctlcluster 9.1 main stop -- -m immediate
Stopped the main postgres process but the writer process was still running:

> ps -ef | grep postgres
postgres 14228 14224  0 14:34 ?        00:00:03 postgres: writer process
wegnet   28059 22011  0 12:48 pts/0    00:00:00 tail -f
/var/log/postgresql/postgresql-9.1-main.log

Also pg_ctlcluster exit status was 0 which is not helpful in such a
situation.

One note: During all this the server was under heavy load recovering
lots of wal files.

Any ideas?

Best,
Juergen



Вложения

В списке pgsql-admin по дате отправления:

Предыдущее
От: Asmus Reinhard
Дата:
Сообщение: Restore after pg_dumpall
Следующее
От: German Becker
Дата:
Сообщение: Re: Disk latency goes up during certaing pediods