I am Running Postgres 8.3 (old I know but management won't let us upgrade). I have a Warm Standby database on another host which is usually in a state of continuous recovery, using PG standby.
I have recently had to re-silver the standby and that's where the problems start.
I created the Standby from a Tarred Hot backup of the Primary and provided WAL logs until up-to-date, the trouble is that I am always getting issues with one of the logs producing an error such as -
2016-05-24 14:03:17 BST LOG: restored log file "000000030000013300000031" from archive
2016-05-24 14:03:17 BST LOG: unexpected pageaddr 133/1F000000 in log file 307, segment 49, offset 0
2016-05-24 14:03:17 BST LOG: redo done at 133/30311CD0
2016-05-24 14:03:17 BST LOG: restored log file "000000030000013300000030" from archive
2016-05-24 14:03:17 BST LOG: selected new timeline ID: 4
2016-05-24 14:03:17 BST LOG: archive recovery complete
2016-05-24 14:03:17 BST LOG: database system is ready
This then starts the Standby up normally. I don't want this though obviously, I want the Standby to be continuously recovering.
Has anybody come across this issue before, is there a workaround/fix, to do with WAL logging frequency/check-pointing ?