ONTAP Discussions

Kahuna usage

saranraj456
6,371 Views

Hi Frndz,

We faced a CPU usage spike due to increase in kahuna domain usage. Its been happened on fronm 14 th july .So,we are bit curious to know what could be reason for the increase in kahuna.

Saran

4 REPLIES 4

madden
6,369 Views

Generally CPU increases because the system is being asked to do more work.  This work could be user work (i.e. your hosts are asking more of the storage system) or system work (i.e. more replication, or tape backups, or deduep jobs, or something else).   Your question was about Kahuna specifically, but because this is the default processing domain it includes many user and system functions and is not specific enough to point toward a cause.  More about processing domains is here: https://kb.netapp.com/support/index?page=content&id=3010150

I'd look to see if the user work requested changed from around this date.  If it didn't, then I'd look to system work and the schedule of things like dedupe and replication.  If you have many volumes with the same and frequent schedules you might be creating spikes of system work that could be easily avoided.   There could be other sources as well and NetApp support should  be able to help if you log a support case.

If your spike occurs weekly on Sunday at 01:00 and lasts maximum 6hrs this is expected as it is the weekly raid scrub verifying that the original data written is still in tact.  This processing yields to user work but will consume a lot of CPU.  Check the Data ONTAP "Storage Management Guide" for more on this process.

Hope that helps.

saranraj456
6,369 Views

Consistently the CPU & kahuna usage is 90% ,from that date. Support guys are not providing proper response.


Saran

madden
6,369 Views

Too bad that support is not responsive.  If you have a support contract with NetApp directly I recommend you to reopen the case and then ask to speak to the duty manager.  The duty manager can help right any wrongs that might have occurred.

saranraj456
6,369 Views

Hi all,

We found the cause with the help of the support, it is a ic_bulk data process ,which is triggered mainly by FCP partner path issue.

We believe its to be software bug since the FCP partner path was exists some long time ago.

Saran

Public