| From: | Michael Stone <mstone+postgres(at)mathom(dot)us> |
|---|---|
| To: | Scott Marlowe <smarlowe(at)g2switchworks(dot)com> |
| Cc: | pgsql-performance(at)postgresql(dot)org |
| Subject: | Re: WAL sync behaviour |
| Date: | 2005-11-10 16:00:55 |
| Message-ID: | 20051110160054.GP9905@mathom.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
On Thu, Nov 10, 2005 at 09:52:38AM -0600, Scott Marlowe wrote:
>Not from what I understood. Ext2 can't guarantee that your data will
>even be there in any form after a crash.
It can if you sync the data. (Which is the whole point of the WAL.)
>I believe only metadata journaling is needed though.
If you don't sync, metadata journaling doesn't do anything to guarantee
that your data will be there, so you're adding no data security in the
non-synchronous-write case. (Which is irrelevant for the WAL.)
What metadata journalling gets you is fast recovery from crashes by
avoiding a fsck. The fsck time is related to the number of files on a
filesystem--so it's generally pretty quick on a WAL partition anyway.
Mike Stone
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Ron Peacetree | 2005-11-10 16:23:23 | Re: Sort performance on large tables |
| Previous Message | Scott Marlowe | 2005-11-10 15:52:38 | Re: WAL sync behaviour |