Oh, so *all* the transactions are being slowed down at that point...What about
CPU IO Wait% at that moment? Could be some other processes stressing the system
out?
Now im thinking about hard disk issues...maybe some "smart" messages?
Have some other hardware to give it a try?
Gerardo
- Me
Hi Gerado,
Thanks for the quick response. We do not appear to have a connection limit
since our application is the only thing talking to the database, the
connections are somewhat limited. We are using about 126 of a max allowed 350
connections. We keep these metrics in a different database, an
After quick reading, im thinking about a couples of chances:
1) You are hitting a connection_limit
2) You are hitting a lock contention (perhaps some other backend is locking the
table and not releasing it)
Who throws the timeout? It is Postgres or your JDBC connector?
My initial blind guess is
Transactions to table, ChangeHistory, have recently become intermittently slow
and is increasing becoming slower.
* No database configuration changes have been made recently
* We have run vacuum analyze
* We have tried backing up and reloading the table (data, indexes, etc)
Some transactions res
