Re: PANIC: replication checkpoint has wrong magic 0 instead of …??

Just moving away to somewhere else

$PG_DATA/pg_logical/replorigin_checkpoint

worked well in my case. :-)

As to be seen in src/backend/replication/logical/origin.c,
without replorigin_checkpoint, PG skips comparison, and copies the other
magic into a new $PG_DATA/pg_logical/replorigin_checkpoint.

On Sun, Jul 19, 2020 at 3:35 PM Nick Rudnick <nick(dot)rudnick(at)gmail(dot)com> wrote:

> Dear community,
>
> after a hardware failure, I get (with PostgreSQL 11.2) output at starting;
> the partition itself appears to be clean:
>
> 2020-07-19 13:08:14.825 GMT [2410] LOG: listening on IPv4 address
> "127.0.0.1", port 5432
> 2020-07-19 13:08:14.838 GMT [2410] LOG: listening on Unix socket
> "/tmp/.s.PGSQL.5432"
> 2020-07-19 13:08:15.041 GMT [2411] LOG: database system was shut down at
> 2020-07-06 08:05:05 GMT
> 2020-07-19 13:08:15.045 GMT [2411] PANIC: replication checkpoint has
> wrong magic 0 instead of 307747550
> 2020-07-19 13:08:15.187 GMT [2410] LOG: startup process (PID 2411) was
> terminated by signal 6: Aborted
> 2020-07-19 13:08:15.187 GMT [2410] LOG: aborting startup due to startup
> process failure
> 2020-07-19 13:08:15.188 GMT [2410] LOG: database system is shut down
>
> A web search wasn't too verbose about this.
>
> Any ideas? Can this be fixed somehow?
>
> Thanks a lot in advance, Nick
>