On Tue, Aug 20, 2013 at 8:17 AM, Sergey Konoplev <gray.ru@gmail.com> wrote:
> Hi all,
>
> My WAL archiving script based on pg_receivexlog reported the following
> error several days ago (just ignore everything before
> 'pg_receivexlog', it's a message my script generates).
>
> Thu Aug 15 18:33:09 MSK 2013 ERROR archive_wal.sh: Problem occured
> during WAL archiving: pg_receivexlog: could not send feedback packet:
> server closed the connection unexpectedly
>
> At the same time postgres reported this error in log:
>
> 2013-08-15 18:32:51 MSK 30945 postgres@[unknown] from [local]
> [vxid:53/0 txid:0] [streaming 2A97/6FA48000] LOG: terminating
> walsender process due to replication timeout
>
> Both pg_receivexlog and postgres run at the same machive,
> pg_receivexlog connects to postgres locally. /var/log/messages has
> absolutely nothing about it. I also have a hot standby on another
> machine connecting to the same master, but there is nothing strange in
> its logs either.
>
> Any thoughts what it was?
It looks like something outside postgres or pg_receivexlog did
terminate the connection. pg_receievexlog noticed it was closed,
whereas postgres waited for the timeout (but probably would've noticed
if it had actually had some other data to send maybe?). Do you have
some iptables connection tracking or something like that which could
be interfering?
--
Magnus Hagander
Me: http://www.hagander.net/
Work: http://www.redpill-linpro.com/