We just migrated our system from Informix/GeneroDB to PostgreSQL 9.03. We are monitoring the situation very closely.
Everything seems to be OK except our concern on the build up of WAL files. You see, we cut off from the previous database around 3.30AM. After that we directed our applications to point to PostgreSQL. We already had a Standby machine A up and running as a Secondary for log shipping and replication. We also setup another machine B just to receive the WAL(s) and store it for us to backup later for recovery purposes. As there are no PostgreSQL running in that machine no pg_archive_cleanup is in effect. Our plan is to backup the WAL(s) every 12 hours and then purge them all. But at the rate the logs are building up at machine B, we would possibly ran out of space very soon.
But this is the point to note. There is very little transaction taking place and the checkpoint_timeout is set to 35minutes. Despite that the WAL logs are getting shipped every minute.
