Unexpected Signal 11

Some people have been seeing OpenNMS die, sometimes frequently, sometimes after a month. In each case, output.log has a Java VM error message that references “Unexpected Signal 11″.

I haven’t seen it on any of my machines, but I have seen it at some client sites. Recent Wrolf Courtney was having the problem quite regularly. He got a tip from Bill Ayres to increase his “ulimit” values.

In opennms.sh change the following line:

ulimit -s 2048

to

ulimit -s 8192

and add the line

ulimit -n 10240

Wrolf said that this solved his problem, and it makes sense to me. I have seen strange “OutOfMemory” and “TooManyOpenFiles” errors when the JVM gets constrained by the O/S, and with RHL 8.0 they have added pam_limits by default, so my guess is that I am not seeing it since I tend to still run on Red Hat Linux 7.3

Anyway, if you are seeing this problem, you may want to check out that change. I have submitted it to CVS HEAD, and Wrolf has updated the FAQ.

General Purpose Poller Updated

Bill Ayres “Gp” script poller has been updated in CVS. Instead of position specific parameters, the hostname and timeout will be passed as options. From the code:

The external script or program will be passed two options: –hostname, the IP address of the host to be polled, and –timeout, the timeout in seconds.