On Fri, Sep 30, 2011 at 1:35 AM, Ian Harding <harding.ian@gmail.com> wrote:
> I updated the firewall rules on a streaming replication standby server
> and thought nothing of it. I later happened to notice on the primary
> that ps aux | grep stream didn't show streaming to that server
> anymore. On the standby that command still showed the wal receiver
> patiently waiting for new data.
>
> I know I broke it, but would anything have eventually happened, or
> would the wal receiver keep patiently waiting as the world passed it
> by?
I guess that walreceiver has been stuck because it could not detect
the termination of connection for a while. By default, keepalive is
enabled on the replication connection from the standby to the master,
but how long it takes for walreceiver to notice the termination of
connection depends on the kernel parameters related to keepalive.
By default settings, it takes about two hours.
Setting keepalives parameters in primary_conninfo would help to fix
such a problem.
http://developer.postgresql.org/pgdocs/postgres/libpq-connect.html
Regards,
--
Fujii Masao
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center