On 04/18/2012 03:12 PM, Rickard, David wrote:
We have a legacy PostgresSQL 7.3.4 db on Solaris
That's not legacy, that's antique. I'm basing my commands on recent versions - if the options differed back then you may have to adjust.
which has recently (as in last 24 hours or so) begun refusing TCP/IP (JDBC) connections from the same server the db is on.
The db is up and running—psql connects;
When you say psql connects do you mean that when you run psql on the same Solaris machine as the server it connects? And if so, what parameters are you using? Will it connect if you explicitly specify a network address (psql -h 127.0.0.1 ...) to make sure it isn't using a socket connection?
I can access the db via pgAdmin III from a PC; I can connect via JDBC from a PC.
However: any JDBC call (whether command-line app or web app) running on the db server fails with this message:
org.postgresql.util.PSQLException: Connection refused. Check that the hostname and port are correct and that the postmaster is accepting TCP/IP connections.
And the postmaster log contains the following message:
FATAL: unsupported frontend protocol
Mind you… I am using the same JDBC jar (pg74.216.jdbc2.jar) and connection URL in code on my PC and the Solaris server.
Our pg_hba.conf file contains the following lines:
local all all password
host all all 127.0.0.1 255.255.255.255 password
host all all 0.0.0.0 0.0.0.0 password
And “tcpip_socket = true” is enabled in postgresql.conf.
Any ideas?
What changed no matter how trivial? Was the machine rebooted or services restarted? Any patches installed - automatically or otherwise? Any config changes? Any possibility the client side has started using something other than 127.0.0.1 (local addresses can be all the way to 127.255.255.255 and I've seen this when a client, for some reason, uses 127.0.0.2 for its address)?
The "unsupported frontend protocol" leads me to believe that the network connection is succeeding but the client-server conversation thereafter is incorrect or absent.
Cheers,
Steve