I have got an issue with setting up PITR on my postgres server: WAL archiving and restoring works quite well, as well as restoring to a specific point in time. The issue however is, that if i do a long replay all the WAL files are extracted from the archive and start piling up in the pg_xlog directory without ever being deleted, this causes the playback to stop with an error once the disk is full. If i then delete all the logs and restart the server the replay continues until the disk is full again. Repeat this procedure a few times and the recovery is completed sucessfully. (the actual working time of the server is just a few minutes)
Since this is already a pain after only a few weeks of history i cannot take a system like this online as the recovery process would take forever...
I'd appreciate any help you can provide.
Ps.: wal_level is set to minimal and archiving is off during recovery