Over 110% untilization

[jdprovine]

I am seeing my management plane spiking over 100% when do a preview and a commit, what could be causing that ? Also can a defrag be run or need to be run on the hard drive?

[BPry]

debug system check-fragment

debug system disk-smart-info

May both be of assistance if you suspect a drive issue.

[jdprovine]

so can these be running during normal business hours? Will it affect performance - you knew I had to ask 🙂

[jdprovine]

The times that the management plane utilization seemed to spike the most so far is when using the ACC, previewing a commit or a commit.

I

[BPry]

@jdprovine,

You can view the smart info during business hours; the fragmentation test takes about 5 minutes and I wouldn't do it during business hours if you can avoid it. I beleive that the only way to actually de-frag the drive is in Maintenance mode.

[jdprovine]

How do you put the drive in maintenance mode?

[Brandon_Wertz]

@jdprovine I believe @BPry was suggesting the entire firewall be put into maintenance mode.  I don't think you can put a single drive in MM.

That being said...What type of HDs does your FW have?  If SSDs I believe the general consensus is to not defrag them as it reduces the lifespan of the drives:

http://www.pcworld.com/article/2047513/fragging-wonderful-the-truth-about-defragging-your-ssd.html?p...

[BPry]

Correct, if memory serves correctly you can Defrag a hard drive in maintenence mode. @Brandon_Weitz is correct. If memory serves correct I believe you have 3020s correct, so they would have non SSD drives unless you have upgraded them yourself.

[jdprovine]

Well I have a PA 5050 - so whatever drive came in it is what I have, But I do have an HA pair I suppose I could fail it over to the secondary and then run a defrag on the primary and then fail back. I am not sure that at this point the issues really point to the hard drive as yet. As mentioned on a previous comment at this time it appears that the issues occur when using the ACC, previewing a commit and doing a commit

[jdprovine]

I had a 3020 in my last job but at my current job we have an HA pair of 5050's

[jdprovine]

@Brandon_Wertz  the link you sent does not work for me

[Brandon_Wertz]

Sorry about that I copied a URL from "page 2" apparently that doesn't work:

http://www.pcworld.com/article/2047513/fragging-wonderful-the-truth-about-defragging-your-ssd.html 

[Brandon_Wertz]

Hrmm...I think there might be something messed up with how the Palo site is coded because I can't get the HTML link to open when clicking on it on this site, but when going to it from my Chrome history or from a Google search it opens just fine.  (See the URLs are the same)

[Brandon_Wertz]

@BPry learned something new, thanks.  Didn't know you can put a single drive in maintenance mode.

[BPry]

@jdprovine,

O okay. Ya I thought I recalled you mentioning 3020 prior so I was going off that assumption. One thing you could try to do is the following when you are actually experiancing a high CPU count. Probably best to log the whole SSH session output.

> show jobs all (check for running jobs)
> set cli pager off (results will be fully expanded, you won't need to hit enter to scroll to the bottom of the results)
> debug management-server memory info (To check for fragmentation counter in ‘arena’ field)
> debug log-receiver statistics (To check whether logging is too high)
> show system software status (to associate running processes with IDs)
> show management-clients
> show system resources follow
Press number ‘1’ and then followed by <shift>H (to show threads). Let it run for a 30 seconds. Hit ‘M’ to sort by memory usage. Let it run for 30 seconds, and press ‘q’ to quit from it.
> show system disk-space

All of this can be run without much load to the CPU so I wouldn't be to worried about running it during the day. If you want to have the cli paging turned back on just remember to run the set cli pager on command; but leaving it off isn't going to cause any ill affects to your device, I actually leave paging disabled on my units so I don't have to scroll through the results.