[pgpool-general: 1823] pgpool_switch_xlog on Amazon S3

David Keeffe david.keeffe at rivustv.com
Wed Jun 5 15:00:28 JST 2013


Hi all

I'm just setting up pgpool replication on Amazon AWS.

I'm sending WAL archives to an S3 bucket, which doesn't appear as a 
directory on the server.
PgSQL allows this by simply requiring a program to be called to perform 
the archiving, similarly for recovery.

However, pgpool_switch_xlog expects a standard folder in which the WAL 
files are written and clearly won't work in this case.

I see two possibilities:
* change my archiving script to create a dummy file in a local 
directory, suitable for pgpool_switch_xlog to check, or
* change my archiving script to call pg_switch_xlog, and then check 
$PGDATA/pg_xlog/archive_status/NNNNN.done in a shell script loop.

The first is probably easier to implement, and won't cost much storage.
The second is more complex but doesn't require any extra files.

Does anyone care to comment?

David K



More information about the pgpool-general mailing list