Gentoo VMWare Workstation Host, sluggish X CPU increase

[volumen1]

So, this is a bit of a longshot, but I'm hoping that someone might have VMWare-specific tuning tips for running VMWare workstation on a Gentoo host. My desktop machine is a pretty decent box, here's the details

Intel(R) Core(TM) i7 CPU 860 @ 2.80GHz
12G RAM
Intel fakeraid RAID controller with mirrored 1.5TB drives

I've been running VMware workstation for a long time, but the sluggishness is finally starting to drive me crazy. The particular VM I'm working with is a P2V conversion, if that makes any difference. Basically, what I notice is that when I start my VM (even if it isn't doing anything) switching between programs (or browser tabs) starts to lag. Just, in general, on the Gentoo host machine, everything seems to slow down. The VM that I'm running is configured for 2G RAM and configured to use one CPU with 2 cores. One thing that might be significant is that the disk is fairly large. It's configured for 209G of disk and is currently using 112G. The virutal disk contents are stored in a single fisk and disk space is not preallocated for the virtual disk.

Anyway, before I start the VM, here is what top looks like:

[wagglet]

Hmm interesting problem.
Perhaps you first need to get some real-time performance data (processes, CPU, I/O, Mem, Slabs etc., over time) that you can put through e.g. GNUplot. In the good old days this meant using sar and family, but collectl seems to offer a more useful solution whilst having minimal performance hit, depending upon configuration. See http://collectl.sourceforge.net/ for more info. Portage has an ebuild available (masked). Also don't forget (as I have done before now!) to document what's happening inside your VM at the same time; what does your VM think is happening? Once you have concrete data from controlled conditions, you can then investigate why you are having this issue (e.g. My system is disk I/O blocked when x, y, z happens).

Prima facie, though, from my own experience, a couple of flags popped up when reading your post:

• P2V configurations have never really worked THAT well for me, though they have worked - now I always start from a 'clean' install, raw disks giving some performance boost (but they come with their own headaches)
  
• From early on I have stuck to static, preallocated virtual disks. Provided you plan your capacity requirements carefully, this shouldn't cause any problems and eliminates the need for VMWare to waste time growing your virtual disk, amongst other things.
  
• Does your VM perhaps have a bloat issue also? My VMs are always stripped as far as possible from unneeded components, apps, services etc. and I generally have one VM for one purpose. Depends on your unique requirements, of course
  

By all means, if you can, create a new 'clean,' static-disk VM and see how that performs - that will give a quick indication of whether it is your VM or VMWare itself that is having the problem, but it won't tell you precisely why this is happening.

That's my two cents' worth anyway...hope it gives you some ideas and you don't feel I've told you how to suck eggs lol. It'll be interesting to see how you get on.

[volumen1]

I appreciate the advice, especially the bit about collecti (as that is new information for me). Interestingly enough, I think I solved this issue yesterday and it was goofy. I believe it was a combination of two things.

1) nvidia-driver had been updated but I hadn't unloaded the nvidia module and restarted X yet, I've notice this always gives me weirdness in my VMs
2) another machine in my network was given a static IP in the DHCP range and it happened to be conflicting with the ip assigned to the VM. I wouldn't have expected this to do anything other than make networking spotty at best, but I think that the vmware network modules were really having a hard time with it?

Anyway, the VM is peppy as all getout now, so I'm going to tweak and test some more.

Thanks much!
_________________
I was born with a freakin' dice bag on my belt.
-- www.howsyournetwork.com