Обсуждение: WAL archive deletion methodology

Hi,

when a warm standby "slave" database is being maintained, what is the
best way to keep the WAL archive files from cluttering up the disk?
According to
http://www.postgresql.org/docs/8.2/interactive/warm-standby.html it
seems to me that one could either initiate periodic backups (stop PITR
slave; pg_start_backup() & rsync & pg_stop_backuo(); start recovery) and
delete all files before the "START WAL LOCATION" mentioned in the
*.backup file, or just keep the PITR slave running and use
pg_controldata to identify the files that are no longer needed (although
I'm not sure how to translate a checkpoint location to a WAL archive
filename).

What's a reliable way to do this?

Thanks,
  Marinos