| From: | "Fabio C(dot) Bon" <bon(at)kernel(dot)net(dot)uy> |
|---|---|
| To: | <pgsql-admin-owner(at)postgresql(dot)org> |
| Subject: | Questions about fsync |
| Date: | 2005-07-29 16:30:46 |
| Message-ID: | 025b01c5945a$dfc1ba70$0501a8c0@fabio |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-admin |
In the context of PostgreSQL 7.4, running over Linux Suse 9.0, I have some
doubts that wanted to evacuate:
In which processes impacts the property fsync?
Which can be the problem to use fsync in false?
When use fsync = false, the effective recording of the data in disk is done
for criteria of the operating system, ¿ Is exact?
Considering fsync = false, checkpoint_timeout=300.
It given a crash of the operating system, after normally db shutdown. At db
startup, there was corruption of data and a pg_resetlog
was done.
It is possible to lose data changed previous to last checkpoint ?
Thanks,
Fabio
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Chris Travers | 2005-07-29 16:50:07 | Re: Indexes on RAM disk = insanity? |
| Previous Message | Tom Lane | 2005-07-29 16:21:02 | Re: Indexes on RAM disk = insanity? |