OM CPU report and alarm inaccurate on new hardware platforms

[dgreg]

Dear NetApp Community,

The current OM CPU alarm and report takes it's value from the ANY1+ counter in Data ONTAP, as does sysstat -x 1.   On multi-processor /  multi-core storage systems such as the FAS6210A storage system which has eight processing cores, the OM CPU threshold is often breached (95%) causing undue concern to customers when OM issues an alarm. ANY1+ exceeds 95% when processes such as deduplication kick in. However when you look at the output of sysstat -M, each individual processor cores isless than 50% utilized, and processing domains are not stretched.

My question is to the OnCommand UM experts - is there a more intelligent way to monitor CPU utilization and produce an alarm or report when a threshold on an individual processor or core is breached? Thus far I can't see how to do this in OnCommand UM 5.0.2 or 5.1, which have great customized reporting capabilities but not for performance counters as far as I can see.

Thanks in advance.

Greg Davis

PSC, ASEAN

[adaikkap · Accepted solution]

Hi Davis,

     The value show in OnCommand for cpu utilization is the value return by snmp which includes all processors. If you would like to have individual processors cpu utilization, pls take a look at the Performance Advisor which gives the breakdown at individual processor level.

Regards

adai

[dgreg]

Hi Adai,

Thanks for your email and reply. Yes I agree, I've now found out how to set up a Threshold and Alarm for per processor CPU utilization in Performance Advisor, which is a much more useful counter to user for multicore / multiprocessor storage controllers.

Cheers,

Greg Davis

NetApp PSC, ASEAN