We have an OpenNMS 1.6.7 server monitoring 272 nodes with 3901 interfaces, although each node has only 1 or 2 interfaces actually monitored. We noticed the 100Mb connection between the poller and database server being pegged so we replaced with a gigabit switch. Now the interface is spiking up to 300-400Mbps. This is the traffic coming from the database server to OpenNMS. I did a tcpdump and virtually all of the traffic is repeated calls of this query:
SELECT * FROM snmpinterface WHERE nodeid = '325'
The actual nodeid varies, but they are repeated a lot. I saw one node query repeated over 100 times in a capture that was about 5 seconds long.
David Hustace suggested disabling thresholding and that made the problem go away.
Environment
Operating System: Linux
Platform: All
Acceptance / Success Criteria
None
Lucidchart Diagrams
Activity
Show:
Benjamin Reed March 4, 2010 at 2:52 PM
the fix was committed to 1.6 revision 6925ca766962751309dd329957e059add756b694
Alejandro Galue February 25, 2010 at 9:39 AM
Fixed. Next versions, 1.6.10 and 1.7.10, will not be suffering this problem.
We have an OpenNMS 1.6.7 server monitoring 272 nodes with 3901 interfaces,
although each node has only 1 or 2 interfaces actually monitored. We
noticed the 100Mb connection between the poller and database server being
pegged so we replaced with a gigabit switch. Now the interface is spiking
up to 300-400Mbps. This is the traffic coming from the database server to
OpenNMS. I did a tcpdump and virtually all of the traffic is repeated
calls of this query:
SELECT * FROM snmpinterface WHERE nodeid = '325'
The actual nodeid varies, but they are repeated a lot. I saw one node
query repeated over 100 times in a capture that was about 5 seconds long.
David Hustace suggested disabling thresholding and that made the problem go away.