Re: 9.1.3 Standby catchup mode - Mailing list pgsql-general

From Fujii Masao
Subject Re: 9.1.3 Standby catchup mode
Date
Msg-id CAHGQGwEFhMWVyL03v-r37q66TskG_DoBu9_eTVqxswbG1cVGqg@mail.gmail.com
Whole thread Raw
In response to 9.1.3 Standby catchup mode  ("hans wulf" <lotu1@gmx.net>)
List pgsql-general
On Fri, Apr 6, 2012 at 1:35 AM, hans wulf <lotu1@gmx.net> wrote:
> I am wondering how the catchup mode of a hot synchron slave server works on 9.1.3 if there is no WAL archive.
>
> Can the slave only request WALs that are still in the xlog directory of the master server? Or does the master
regeneratesome kind of fake log for the catchup mode? 

No. If the WAL file which the standby requests doesn't exist in the
pg_xlog directory
of the master, replication just fails. In this case, you need to take
a fresh base backup and
start the standby from that backup.

> E.g. in case of a slave failure I could use a weekly backup and let the catchup mode do the rest? Or does that only
workif you use WAL archive? 

Or increase wal_keep_segments to high so that all WAL files which the
standby requests
are guaranteed to exist in the pg_xlog directory of the master.

Regards,

--
Fujii Masao

pgsql-general by date:

Previous
From: hamann.w@t-online.de
Date:
Subject: pltcl and modules
Next
From: 乔志强
Date:
Subject: Re: [streaming replication] 9.1.3 streaming replication bug ?