Hi all,
I'm hoping someone could help me with this issue, not sure if the
general mailing list is the right place so apologies if this needs to be
directed else where.
We have several clients who are experiencing disconnects to our hosted
postgres server. From the server side we simply see a "could not receive
data from client: Connection reset by peer" error.
On the client side we generally see a "could not receive data from
server: The system cannot find the path specified.".
Googling around the indication seems to be that a firewall or router in
between the client and the server may be dropping the connection
considering it inactive. We have setup keep alives (currently every 5
seconds as a test) and that has improved the situation but as our client
application is a single threaded application and it is possible for it
to be processing data and thus not respond the connection may still be
deemed inactive.
My first question is, if it is the keep alive itself (currently set to
10 retries every 5 seconds) decides that the connection is dead, would
we see a different message in the logs?
My second question is, does anyone know what the most likely culprit is
and been able to change the configuration there?
This is on a Postgres 8.4 server running on CentOS but we've also
experienced this in relation to a Postgres 9.1 server
Kindest Regards,
Bastiaan Olij