"Elton Borssoi" <eltonbor@gmail.com> writes:
> The test I made and got this message (PANIC: could not locate a valid
> checkpoint record) is
> 1) everything is stoped
> 2) start pg1
> 3) restore dump on pg1
> 4) stop pg1
> 5) configure pg1 for archiving on /home/postgres/archives
> 4) copy data dir to pg2
> 5) start pg1 and create some tables to generate data volume
> 6) configure pg2 for recovery mode coping from /home/postgres/archives
> 7) start pg2 (this start is fine, everything was recovered)
> 8) stop pg2
> 9) generate more data on pg1 (there are more 2 or 3 archives file)
> 10) start pg2 on recovery mode again and got this message
> PANIC: could not locate a valid checkpoint record
You can't do that. Once pg2 is started it is no longer a slave to pg1;
it's started to develop its own transactional history branching off from
what pg1 did.
regards, tom lane