I have 2 identical machines, (Ubuntu 10.04 LTS running 9.1.2 from Martin Pitt's PPA) setup with streaming asynchronous replication.
Recently I had to take the standby down for some extended maintenance, and when it came back up it was waiting for a WAL segment that had since been removed from the primary. I then shutdown the server, took a new base backup of the primary and copied it over (as per the directions in the streaming wiki) and cleared out the wal segments in the pg_xlog directory of the standby and tried to start it again, but now its not accepting connections and is looking for that wal file in the startup process:
postgres 10090 10089 0 12:38 ? 00:00:00 postgres: startup process waiting for 0000000100000052000000E9
However, segment 52000000E9 is no longer on the primary, which is about a week ahead:
postgres@pg1:~$ psql -c "SELECT pg_current_xlog_location()"
pg_current_xlog_location
--------------------------
58/8E1B8D98
So, how do I get the standby to catch up to "catch up" to the primary?
-Adam Cornett
