The following bug has been logged online:
Bug reference: 4854
Logged by: Keith Pierno
Email address: kpierno@lulu.com
PostgreSQL version: 8.3.6
Operating system: Red Hat Enterprise Linux AS release 4 (Nahant Update 6)
PPC64
Description: Problems with replaying WAL files on Warm Standby
Details:
We experienced a failure of our primary database server, after fixing the
hardware issue I attempted to turn the old primary into a warm standby
slave. I used a hotbackup from the new primary server and restored the files
to the slave. I then removed all references to old WAL files and used rsync
to copy the appropriate WAL files from our archive location. During the log
playback I received the following errors.
2009-06-15 13:25:11 EDT <::> LOG: restored log file
"00000004000007460000007F" from archive
2009-06-15 13:25:11 EDT <::> LOG: restored log file
"000000040000074600000080" from archive
2009-06-15 13:25:11 EDT <::> LOG: restored log file
"000000040000074600000081" from archive
2009-06-15 13:25:11 EDT <::> LOG: restored log file
"000000040000074600000082" from archive
2009-06-15 13:25:12 EDT <::> LOG: restored log file
"000000040000074600000083" from archive
2009-06-15 13:25:12 EDT <::> LOG: restored log file
"000000040000074600000084" from archive
2009-06-15 13:25:12 EDT <::> PANIC: unexpected timeline ID 6 (after 4) in
checkpoint record
2009-06-15 13:25:12 EDT <::> CONTEXT: xlog redo checkpoint: redo
746/84FFF010; tli 6; xid 0/50577196; oid 31395572; multi 6542; offset 13124;
shutdown
2009-06-15 13:25:13 EDT <::> LOG: startup process (PID 27836) was
terminated by signal 6: Aborted
2009-06-15 13:25:13 EDT <::> LOG: aborting startup due to startup process
failure
This backup was created at 0205 EDT and the WAL in question was from 0743
EDT.