From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Vince Vielhaber <vev(at)michvhf(dot)com> |
Cc: | pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: why? |
Date: | 2002-01-11 16:42:31 |
Message-ID: | 29350.1010767351@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Vince Vielhaber <vev(at)michvhf(dot)com> writes:
> Why does postgresql have to die if the disk fills up? Can't it just
> go into an idle state and complain or something?
AFAIR the curl-up-and-die response only occurs if we fail to obtain
space for XLOG or CLOG files; failure to extend ordinary data files
isn't fatal.
I don't really see a way around stopping the system when XLOG or CLOG
is broken. With transaction support not working you can't do anything.
A bright spot is that in 7.2, since we generally recycle rather than
delete/recreate XLOG files, adding space for XLOG is a rare event.
CLOG doesn't grow very fast either (2 bits per transaction). So
you should be more likely to see out-of-space reflected as a user data
file extension failure before you run into XLOG/CLOG trouble.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2002-01-11 16:42:43 | Re: again on index usage |
Previous Message | Don Baccus | 2002-01-11 16:41:11 | Re: again on index usage |