Re: PITR archive_timeout Command - Mailing list pgsql-admin

From mattcwood
Subject Re: PITR archive_timeout Command
Date
Msg-id 24923536.post@talk.nabble.com
Whole thread Raw
In response to Re: PITR archive_timeout Command  (Ian Lea <ian.lea@gmail.com>)
List pgsql-admin

Ian Lea wrote:
>
> The advantage of setting it high is that you'll use less disk space
> and have fewer files to archive.
>
> The disadvantage of setting it high is that you might lose more data.
>

This is not *entirely* true.  The archive_timeout setting only indicates the
time at which a new WAL archive is forced, no matther how much data has been
written.  If a WAL file fills up before the timeout is reached, it will fire
the archive_command immediately and the timeout "timer" is reset.

For example, we have a primary database instance where we see the
archive_command fire every 1-2 minutes during peak, but our archive_timeout
is set to 15 minutes.  This ensures that the secondary database instance is
getting updated even when there's not enough write traffic to fill up a WAL
file.

--Matthew Wood



--
View this message in context: http://www.nabble.com/PITR-archive_timeout-Command-tp24788681p24923536.html
Sent from the PostgreSQL - admin mailing list archive at Nabble.com.


pgsql-admin by date:

Previous
From: "Kevin Grittner"
Date:
Subject: Re: replay WAL segments without a "base backup" ?
Next
From: Kevin Kempter
Date:
Subject: insert a backslash into a bytea field?