We have an adapter notification that polls very frequently (approximately 3 times a minute). This morning we noticed the integration was not running (this is a critical part of the business infrastructure and gets a lot of attention) and got an error in the IS logs saying “The connection is closed”.
What causes this? If the connection times out for a network reason, does it self-heal or do we have to manually go and re-enable the connection?
this looks like a stale connection problem. happened with me in all the connections which have frequent/too many db interactions.
change ‘Minimum Pool Size’ to 0 in your connection properties. hope this works.
Let me know if it still has issues.
Thanks Anjni. Are you saying this would give a “This connection is closed” error? I’m not aware of a stale connection causing this?
Either way, the pool size is already 0.
What I’m thinking is that a network outage caused the connection to disable itself, and it doesn’t re-enable, which is why I’m asking whether connections can self-heal.
Yes I used to get these errors in my production envmnt and after changing the value to 0 it has stopped.
Usually the behavior with network outages what I have seen is it connects again when the network is back. As my envmnt is on latest fixes so this might also be a bug. :roll:
I suspect the high volume of transaction caused this issue. Please try to adjust watt.art.tmgr.timeout in extended setting. It signifies seconds before transaction timeout.
Hi Grant,
Is it a one time issue or keep repeating. If it one time, then kindly don’t do any changes in Prod. In case,if it is repeating kindly share more details to better suggest.
The network going down is the reason the connection timed out; what I didn’t understand was why the connection didn’t re-establish correctly upon the return of the network connectivity.
The DB is very quick, so there won’t be problems there, and we throttle the select to be at most 200 rows at a time. Notification interval is set to provide a 10s gap between processing runs.
What is the minimum connection pool set on the JDBC connectin properties 0 or some thing else?
Some times even though the DB comes back there might be stale connections not released from the pool or network latency issue and this would cause mess up especially JDBC notifications and also what JDBC Adapter6.5 fix level are you on latest I assume?