Gembudders,When analyzing larger datasets with gdplot & gddiag, we notice spikes in our kernel context switches (see attached graph) to what may be unhealthy levels, while CPU load is negligible. Does anyone else see this type of behavior? Is there anything that can be done to prevent this from happening? Our Nagios system sends alerts anytime the number of kernel context switches goes above 150000 (warning) and 200000 (critical) per second.
We're running on CentOS 6.6, Gempak 7.2 built from source, 32 GB of memory, Xeon X5672 3.2 Ghz processor (x8),
Any help is greatly appreciated. Thanks, Brendon -- Brendon Hoch, M.S. Technology Manager Judd Gregg Meteorology Institute MSC 48, Boyd Hall 319 Plymouth State University Plymouth, NH 03264 (603)535-2818 Fax: (603)535-2723
Attachment:
kcsgraph.PNG
Description: PNG image
gembud archives: