2013-10-18 05:10 AM
we have a fas3210 with PAM and data ontap 8.1.3
i have noticed very high CPU util going up and down all the time... i seached the bud id section and found somethings for kahuna domain and block reclamation and upgraded to 8.1.3 but the problem is not solved yet.
is there anyboady else facing this problem and if yes which is the solution on this?
2013-10-18 05:45 AM
There is a whole lot of reason why the CPU could be busy, and the answer usually comes from a perfstat and a call to NetApp Support.
Deduplication is one of them, as is SnapMirror especially when using high speed links, block reclamation as you said, if you deleted a large amount of snapshots for example, volume realloc, etc etc...
You also need to avoid using the single number given by the filer as CPU usage, but do a sysstat -M to see what each CPU is using, it might be very different and not that bad finally.
Do you observe any latency / slowness on the clients side ?
2013-10-18 07:20 AM
Hi Yann thanks for the answer...
i am just trying to identify the cause... the thing is that after the upgrade one of the controllers has the problem which is the secondary filer for the snapvault and i read to a thread that there might be a bug in the snapvault...
also i saw 2 idle processes consuming 94 and 95% of the CPU
2013-12-05 03:33 AM
Did you find out the root cause.
I dont think think 8.1.3 has any bug that could cause CPU to spike (we faced this issue in 8.1.2 though)
Does a high CPU in your case brings along latency as well for the volumes causing slowness in reads or writes?High CPU is by itself not a indicator for performance issue unless it is accompanied by latency.
Also dont rely on systat -m always as it will give you an aggregate % times used for all CPU's.A more reliable paramter should be "priv set diag;systat -M 1"