| Lists: | pgsql-general |
|---|
| From: | Kevin Kempter <kevin(at)kevinkempterllc(dot)com> |
|---|---|
| To: | pgsql-general(at)postgresql(dot)org |
| Subject: | DRDB risk factors? |
| Date: | 2007-05-30 21:31:36 |
| Message-ID: | 200705301531.36535.kevin@kevinkempterllc.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Lists: | pgsql-general |
Hi List ;
per considering DRDB as a replication solution in a failed master node
scenario, is there a risk of loosing not only in-flight transactions but alos
un-sync'd buffer-pool dirty pages?
If so, how might I minimize this risk ?
Thanks in advance...
| From: | Ben <bench(at)silentmedia(dot)com> |
|---|---|
| To: | Kevin Kempter <kevin(at)kevinkempterllc(dot)com> |
| Cc: | pgsql-general(at)postgresql(dot)org |
| Subject: | Re: DRDB risk factors? |
| Date: | 2007-05-30 22:20:18 |
| Message-ID: | Pine.LNX.4.64.0705301519300.27105@localhost.localdomain |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Lists: | pgsql-general |
If it's been fsync'd to a DRDB device, it's been fsync'd to both systems.
I think that answers your question...
On Wed, 30 May 2007, Kevin Kempter wrote:
> Hi List ;
>
> per considering DRDB as a replication solution in a failed master node
> scenario, is there a risk of loosing not only in-flight transactions but alos
> un-sync'd buffer-pool dirty pages?
>
> If so, how might I minimize this risk ?
>
> Thanks in advance...
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: if posting/reading through Usenet, please send an appropriate
> subscribe-nomail command to majordomo(at)postgresql(dot)org so that your
> message can get through to the mailing list cleanly
>
| From: | Hannes Dorbath <light(at)theendofthetunnel(dot)de> |
|---|---|
| To: | Kevin Kempter <kevin(at)kevinkempterllc(dot)com> |
| Subject: | Re: DRDB risk factors? |
| Date: | 2007-05-31 07:31:58 |
| Message-ID: | 465E79EE.7040908@theendofthetunnel.de |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Lists: | pgsql-general |
On 30.05.2007 23:31, Kevin Kempter wrote:
> per considering DRDB as a replication solution in a failed master node
> scenario, is there a risk of loosing not only in-flight transactions but alos
> un-sync'd buffer-pool dirty pages?
>
> If so, how might I minimize this risk ?
If you have DRBD using protocol C and fsync enabled in PG you don't lose
a single commit. That is true for any ACID conform application.
Be sure to you have your hardware right. Running DRBD on fsync/fua lying
hardware is committing suicide. I can't stress that enough.
--
Regards,
Hannes Dorbath