ONTAP Discussions
ONTAP Discussions
I noticed this morning, between 4:00am-4:30am, one of my controllers spiked to 100% CPU Utilization. Screenshot attached.
I drilled down to each volume to see if any of them where running high IOPs during the same time, and didnt find anything unusual. Most of my snapshots and snapmirror happens between 8:00pm-midnight. Nothing else I can think of runs between 4:00am-4:30am. What else can I look at to determine why this controller hit 100%???
Solved! See The Solution
often we see one cpu pegged but other cpus are free for wafl_exempt processes for user access... if you run "sysstat -M 1" in advanced/diag mode you can see all cpus and cpu domains... I think asis runs in kahuna (please correct me someone in engineering if not) which will use one cpu.
Are you running SIS on any of your volumes?
Does this happen every day?
Yes. I do have deduplication scheduled on about 7 NFS volumes. I did notice a couple of them run longer than others, for some reason. Is Dedup (sis) actually that CPU intensive? Thanks for your response.
Dale
often we see one cpu pegged but other cpus are free for wafl_exempt processes for user access... if you run "sysstat -M 1" in advanced/diag mode you can see all cpus and cpu domains... I think asis runs in kahuna (please correct me someone in engineering if not) which will use one cpu.
Thank you Scott for reminding me about the sysstat –M command. Next time I have a spike, I will look into it.
I have also reorganized my deduplication schedule. I increased the number of days (3 time a week, instead of 2) and spread them out a bit.
Dale
Getting a perfstat during the spike will help find the bottleneck too, but sounds like you are managing it by changing asis schedules.
Typos Sent on Blackberry Wireless