| From: | ITAGAKI Takahiro <itagaki(dot)takahiro(at)oss(dot)ntt(dot)co(dot)jp> |
|---|---|
| To: | "Jignesh K(dot) Shah" <J(dot)K(dot)Shah(at)Sun(dot)COM> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: LDC - Load Distributed Checkpoints with PG8.3b2 on Solaris |
| Date: | 2007-11-14 08:55:47 |
| Message-ID: | 20071114174139.30F3.ITAGAKI.TAKAHIRO@oss.ntt.co.jp |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
"Jignesh K. Shah" <J(dot)K(dot)Shah(at)Sun(dot)COM> wrote:
> I am running tests with PG8.3b2 on Solaris 10 8/07 and I still see IO
> flood when checkpoint happens.
Are there any i/o tuning knobs in Solaris? LDC in 8.3 expects writing
activity in kernel is strong enough to keep dirty pages in kernel
in a small number. i.e, fsync should be cheap.
Larger shared_buffers, checkpoint_timeout, checkpoint_segments and
checkpoint_completion_target make checkpoints more smooth.
But there is a trade-off between smoothness and recovery time.
Stronger writer processes in kernel make checkpoints more smooth.
But there is a trade-off between smoothness and throughput.
Regards,
---
ITAGAKI Takahiro
NTT Open Source Software Center
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Magne Mæhre | 2007-11-14 09:34:51 | Spinlock backoff algorithm |
| Previous Message | Magnus Hagander | 2007-11-14 08:51:53 | Re: Simplifying Text Search |