UHm - WHM or virtuosso CPU usage monitoring?

**Kaiberie** · 03-01-2006, 04:05 PM

Ok - now I'm REALLY confused.

WHM says I'm in the red for resources - Virtuosso says I'm not, and I'm really low on usage - which do I go by?

The reason I ask is cause my sites aren't especially slow (and are faster since upgrade, as expected) but if that's my defining way of choosing to move, (server load/CPU usage) how can I tell which is right and which isn't?

Thanks for any advice you can give

Kai

**gerilya** · 03-01-2006, 05:14 PM

Why wouldn't you set up your own resource monitoring tool then?

**Kaiberie** · 03-01-2006, 05:17 PM

Could you reccomend one please? I'm a 'lil new at this stuff (can you tell?)

Kai

**the_ancient** · 03-01-2006, 05:38 PM

I dont know much about the VPS's. but maybe Viruozzo is monitoring the whole box, and WHM is telling you your little piece of it?

Or mayvbe WHM is just on crack?????? If your sure Virtuzzo is only looking at your piece of the pie, I would trust it wellllllllllllllll before I trusted whm

**piper** · 03-01-2006, 05:43 PM

I'd like to know this as well. Looking forward to replies.

**tank** · 03-01-2006, 09:38 PM

Log in as root to SSH and type "top"

that will give you an accurate pisture of what your server load is.

**Gwaihir** · 03-02-2006, 06:34 AM

Trust Virtuozzo's readings over WHM's.

WHM wasn't specifically meant to run on a VPS. It expects the available amount of CPU-power, memory and such to be fixed; a hard, hardware based limit. On a VPS those limits are sliding, as you're allowed to tap into the unused resources of other VPSses. Must confuse the crap out of WHM.

Virtuozzo on the other hand is the software that makes those arrangements, so its load measuring tools should be made to cope with it.

**gerilya** · 03-02-2006, 10:16 AM

Originally Posted by Kaiberie

Originally Posted by gerilya

Why wouldn't you set up your own resource monitoring tool then?

Could you reccomend one please? I'm a 'lil new at this stuff (can you tell?)

Try sar (bsdsar for FreeBSD) command first. If you can't find sar on your system, install sysstat rpm.
After you get familiar with it, set it up to be executed from cron each 10mins.

**Kaiberie** · 03-02-2006, 10:40 AM

I'm using top now - what its showing me is far different from either of them.

My average load is 1.1 - it sometimes spikes to 2.3 - 3, but I know what's causing that and we're addressing it. But cpu useage is only 4%. So what should I worry about, load or CPU usage? Or both, or neither? Or?

The other thing I've noticed is that when WHM is crashing, load spikes to 18 or 19, just to get the bloody CPRSVD or whatever its called restarted. And then the server goes offline for a couple of minutes.

What really bothers me is that its fine whilst running - its not overly slow (in fact its faster than either of the servers I've been on to date) but its also prone to locking up.

Sorry, I'm a noob - just throw things at me if you like

Kai

**tank** · 03-02-2006, 03:10 PM

Load is a function of CPU usage. when you run top you should have a list of services that shows a percentage of CPU usage. For instance there is a mysql service that may be at 2 or something. what service is showing the highest reading?

**Ron** · 03-02-2006, 10:26 PM

Load is NOT just a function of CPU usage. Load is the average runnable-queue depth.

Processes waiting for Disk I/O are on the runnable queue, therefore add to the queue depth. My guess, from his description of 4% CPU with a server load of 1 sounds like his VPS is I/O bound to me.

That I/O may be from straight disk data read/writes, or it could be caused by paging memory to disk. I can't tell from my chair over here.

The top command may give some clue to this; at the top there are numbers like this:
Cpu(s): 6.9% us, 2.0% sy, 0.0% ni, 16.4% id, 74.7% wa, 0.0% hi, 0.0% si

See that 74% wa? That's I/O wait. The CPU is actually idle during I/O wait, the data shows up in this column when there is nothing for the CPU to do but sit around and wait for the I/O to be completed. So the total Idle for this machine is 91%+ while the server load is 9+ because there are so many processes sitting in the wait I/O queue.

Also there is a line about Swap; that's how much disk space has been allocated for swapping data in and out of memory. If the "used" number is high, memory usage would be a good place to start to track down the problem.

**tank** · 03-03-2006, 12:29 PM

Originally Posted by Ron

Load is NOT just a function of CPU usage. Load is the average runnable-queue depth.

yes. i realize I didn't say that explicitly but that's what I was implying. If we can nail down what service is the resource og it would help remendously. A print out of the top report suring high load would be really helpful.

**Kaiberie** · 03-05-2006, 02:23 PM

Actually, I tracked it down - I've altered the load warning setting, turned off logging and backup 24/7 (dunno why it was doing that, its now back to the least used hours on the server) and I'm working on optimising the settings to actually BE useful for a cpanel/VPS install.

Top gave me quite an eye opening look at things though - far and away, I'm going to have to track down why my MYsql spikes so dramatically, so I'm also working on that.

I also found that setting my 'extra cpus' setting in one of the setting places to 2, and upping the load warning to 5 helped tremendously - doesn't lock up quite so often now.

NOW all I have to do is get up the courage to play with the apache settings like it suggests and pray I don't accidentally kill the VPS again

Thanks again for your help!

Kai

LinkBack

Thread Tools

Display

UHm - WHM or virtuosso CPU usage monitoring?

Bookmarks

Bookmarks

Posting Permissions