How to determine historic CPU spikes

[LIGISTX]

I am not sure how I would try and figure out what this is, but it looks like there is an extremely repeatable process running. This may be 100% normal, but it just seems way to repetitive and I am not sure what would be causing it. Likely some system process, but just wondering what it may be, my box has no users accessing it 99% of the time.

Going back on the graph shows its pretty consistent.

 

[Chris Moore]

There must be some process, perhaps something you have scheduled, but there shouldn't be much utilization if you are not doing something.
For example:

However, if I go back to when I ran my last backup, it looks much different:

 

[Chris Moore]

As for figuring out what it is, you could try opening a SSH terminal to the server and running top so that you can see what process is using the cycles. It will give you a list like this that constantly updates:

 

[LIGISTX]

There must be some process, perhaps something you have scheduled, but there shouldn't be much utilization if you are not doing something.
For example:
View attachment 25283

However, if I go back to when I ran my last backup, it looks much different:

View attachment 25284

I agree, the only thing I can think of is ZFS snapshots. I don't have any cron jobs running (as I could never get the few I wanted to work to actually work in 11.1U4), no rsync jobs, no thing scheduled in the webUI, and my jails don't have anything either.

I do have ddns updating, but I can't see that being the cause.

Looks like it is related to disc usage tho... hmm.

 

[LIGISTX]

Is it possible my snapshots are a bit too aggressive? I didn't think they really use any resources if nothing has changed.

 

[LIGISTX]

Ah, its my plex library scan.... I guess that makes sense. I am shocked at how much of a CPU hit that is creating.

 

[Chris Moore]

I agree, the only thing I can think of is ZFS snapshots. I don't have any cron jobs running (as I could never get the few I wanted to work to actually work in 11.1U4), no rsync jobs, no thing scheduled in the webUI, and my jails don't have anything either.

I do have ddns updating, but I can't see that being the cause.

Looks like it is related to disc usage tho... hmm.

Are you doing a snapshot every 30 minutes?

 

[Chris Moore]

Is it possible my snapshots are a bit too aggressive? I didn't think they really use any resources if nothing has changed.

No, that looks fine. With my schedule, I have around 86 at any given time. They do take a very small amount of storage to keep the pointers and the amount of data for the change, but it is not much.

 

[Chris Moore]

Ah, its my plex library scan.... I guess that makes sense. I am shocked at how much of a CPU hit that is creating.

You might want to scan less frequently. I only scan once an hour but even that is probably more often than I need because my library only changes when I manually add something.

 

[Chris Moore]

My Nas: 11.1-U4 - -HPE Proliant ML10 Gen 9 backbone

What CPU are you rocking in that?

 

[LIGISTX]

You might want to scan less frequently. I only scan once an hour but even that is probably more often than I need because my library only changes when I manually add something.

That’s true, it only changes when I add as well. One more reason I want to switch to running plex in a VM, it can’t watch folders for changes correctly in FreeBSD, but I think it should be able to in Debian.

But I agree, until then maybe I should dial that back a touch.


Sent from my iPhone using Tapatalk

 

[Chris Moore]

That’s true, it only changes when I add as well. One more reason I want to switch to running plex in a VM, it can’t watch folders for changes correctly in FreeBSD, but I think it should be able to in Debian.

Sure. If you add media to the FreeNAS it should pick it up on the next scheduled scan or you can manually tell it to scan right then. If it isn't picking up new media, something may be configured badly.

 

[LIGISTX]

What CPU are you rocking in that?

i3, whatever the default in that config was. I am blanking on it at the moment.


Sent from my iPhone using Tapatalk

 

[Chris Moore]

i3, whatever the default in that config was. I am blanking on it at the moment.

This is just conversational, not really important. They had two options as I recall, the Xeon was quad core and the i3 was dual core, so that spike looks higher on the graph because each core is having to work harder since there are only two. If you had the quad core, the graph would flatten out because the work would be distributed across more cores.

 

[LIGISTX]

This is just conversational, not really important. They had two options as I recall, the Xeon was quad core and the i3 was dual core, so that spike looks higher on the graph because each core is having to work harder since there are only two. If you had the quad core, the graph would flatten out because the work would be distributed across more cores.

I agree, still surprised at the load on a dual core with HT. Sure, it’s no powerhouse, but still.

O well, it is what it is.


Sent from my iPhone using Tapatalk