Starting monitoring on the beta cluster, Nagios is complaining about too many processes on the compute nodes. By default, Nagios NRPE warns at 150 and reports critical at 200. Each of these nodes has >200 total processes.

On node1, 244 are kernel processes such as kworker (100), ksoftirqd (24), cpuhp, migration, watchdog. For these big nodes,

***** Nagios *****

Notification Type: PROBLEM

Service: Total processes

Host: workbench-node2

Address: 141.142.210.100

State: CRITICAL

Date/Time: Sat Oct 8 19:45:26 UTC 2016

Additional Info:

PROCS CRITICAL: 281 processes

Temporary workaround: disable total_proc monitoring, which probably isn't good.