From: | Vick Khera <vivek(at)khera(dot)org> |
---|---|
To: | pgsql-general <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: tuning postgresql writes to disk |
Date: | 2011-02-09 14:39:09 |
Message-ID: | AANLkTikAExkhDsRFgNhGc+SUL0pHkJpyg+utgczx3LzL@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Mon, Feb 7, 2011 at 7:43 PM, Vinubalaji Gopal <vinubalaji(at)gmail(dot)com>wrote:
> Thank you. I will try to run some performance tests using the async
> commit option. Is there an easy way to find the lost transactions or
> does it have to be handled by the application?
>
By definition, your application has to be able to detect it. If the DB were
able find them, they wouldn't be lost.
Personally, I turn off synchronous commit on a per-connection basis based on
what my application is doing. When I insert or update data, I need to ensure
it is there; when I move data around or do batch processing on it, it is
usually a restartable operation.
From | Date | Subject | |
---|---|---|---|
Next Message | Adrian Klaver | 2011-02-09 14:44:25 | Re: Possible Bug |
Previous Message | Ray Stell | 2011-02-09 14:12:50 | Re: PostgreSQL 9.0 Streaming Replication Configuration |