> unix_socket_directories Is set to the default one which is /temp Before the minor upgrade it is using the correct location. However after upgrade when I started Postgres using pg_ctl start, logs reported as listening on socket at /tmp. So the server started on correct socket. However when I tried using psql, it is checking a completed different socket location /var/run/postgresql/.s.PGSQL.5432 Thanks and Regards, J. Teja Sri. > On Jun 28, 2022, at 1:26 PM, Tom Lane <tgl@xxxxxxxxxxxxx> wrote: > > Teja Jakkidi <teja.jakkidi05@xxxxxxxxx> writes: >> Cluster started and I was able to connect remotely with no issues. However when I tried using psql, I got below error: >> Connection to server on socket “/var/run/postgresql/.s.PGSQL.5432” failed. >> But my postgres has been using socket file in /tmp location before upgrade. I am not sure why after upgrade psql is searching at a different allocation for socket. >> Can someone help me here? > > Some distributions configure the code with a different default socket > location because they don't like putting the socket in /tmp. It's > best to use a server and libraries all built by the same source so > that you don't have to fight with decisions like that. If you have > to, though, it's possible to force the server to put its socket > where you want (see unix_socket_directories). > > regards, tom lane
