Hello All,
We have a postgres setup on solaris 10 with sun cluster for HA purposes.
2 nodes are configured in the cluster in active-passive mode with
pg_data stored on external storage. Everything is working as expected
however, when we either switch the resource group from one node to other
or rg restart on primary node, the apps fails with "An I/O error
occurred while sending to the backend." and doesn't recover back from db
failover. All queries to the db give the above error after resource
group restart.
Our app uses resin container db pooling, with following HA parameters
set. With same settings, the app recovers just fine with database
configured in non-cluster mode i.e. no sun cluster setup etc.
<database>
<jndi-name>jdbc/nbbsDB</jndi-name>
<driver type="org.postgresql.Driver">
<url>jdbc:postgresql://db-vip:5432/appdbname</url>
<user>appusr</user>
<password>apppass</password>
</driver>
<max-connections>100</max-connections>
<max-idle-time>5m</max-idle-time>
<max-active-time>6h</max-active-time>
<max-pool-time>24h</max-pool-time>
<connection-wait-time>30s</connection-wait-time>
<max-overflow-connections>0</max-overflow-connections>
<ping-table>pingtable</ping-table>
<ping>true</ping>
<ping-interval>60s</ping-interval>
<prepared-statement-cache-size>10</prepared-statement-cache-size>
<spy>false</spy>
</database>
Any pointers to debug this futher is greatly appreciated. We are running
postgres 8.2.4.
Other thing i noticed in pg_log/server.logs is that whenever i restart
postgres i get below error, when there is no other postgres running on
5432.
"LOG: could not bind IPv6 socket: Cannot assign requested address
HINT: Is another postmaster already running on port 5432? If not, wait
a few seconds and retry."
Thanks,
Stalin