Re: [SLUG] VMWare hates me -- guest processes hang

From: Eben King (eben01@verizon.net)
Date: Mon Jul 31 2006 - 13:29:19 EDT


On Mon, 31 Jul 2006, Jason Boxman wrote:

> Eben King wrote:
>> On Mon, 31 Jul 2006, Jason Boxman wrote:
>>
>>> On Monday 31 July 2006 00:18, Eben King wrote:
>>> <snip>
>>>> That indicates to me that the guest OS is trying to invoke a screen
>>>> saver, and VMware doesn't support the method it's using. If there's a
>>>> boot option to disable it, use it. It might be part of some power-saving
>>>> package (think ACPI).
>>>
>>> I finally came across a thread that sounds similar to that, but the issue
>>> was the host OS CPU was being throttled down.
>>
>> Yeah, I think you should disable ACPI from inside the guest OS. The best
>> situation is that it sees the CPU state accurately, and its instructions to
>> the OS are heeded. But the host OS will always have at least as accurate a
>> picture, and its instructions are sure to be heeded.
>
> I thought about it, but my laptop is a P4-based Celeron and doesn't support
> any MHz throttling.

All the more reason to disable it in the guest OS -- if it tries to lower
the CPU speed, VMware may try and wait, hang, or crash trying.

>>> I did finally complete a Dapper text-only install, so it seems I may have
>>> not been waiting long enough. It took two hours. After, even with VMWare
>>> tools, my X session would 'hang' for five to thirty seconds on occasion
>>> with the host OS idle and the guest doing simple things like opening
>>> Konsole windows. I think my laptop is simply too slow, although I had no
>>> issues doing this ages ago with VMWare on a much slower laptop with less
>>> RAM and VMWare 2.x or 3.x. Oh well.
>>
>> Try running "top" in your guest OS to see how much swap it's using. It may
>> be thrashing. Actually I think it might be more efficient to just give
>> VMware more RAM, and skip swap entirely.
>
> That was my first thought, but my favorite part is that wasn't happening at
> all. I was at most 12M in swap.

Out of 13M, that's probably bad; but out of 500M, I wouldn't worry. FWIW, I
did run my laptop for a time netbooted, with no swap, and with the drive
spun down. So no swap is possible, although there are analyses indicating
it's suboptimal. I don't know how that applies to OSes on emulated hardware
though.

> Generally my CPU was always pegged in the host at 100%.

CPU time accounting inside VMware is wonky.

> `vmstat` reported nearly no disk access and never any I/O
> wait at all, just sys pegging the guest. I never could pin it on anything
> specifically as the host was idle. Ordinarily I'd suspect DMA was off in
> either the host or the guest, except it wasn't.

I get small writes, every second or two, with Linux 2.6.15.1 on ext3 * 5.
I think it's somebody updating the journal on mounted partitions with
something like "2006-07-31 13:23:50 no transactions".

-- 
-eben      QebWenE01R@vTerYizUonI.nOetP      http://royalty.no-ip.org:81
CAPRICORN:  The stars say you're an exciting and wonderful person... but
you know they're lying.  If I were you, I'd lock my doors and windows
and never never never never never leave my house again.  -- Weird Al
-----------------------------------------------------------------------
This list is provided as an unmoderated internet service by Networked
Knowledge Systems (NKS).  Views and opinions expressed in messages
posted are those of the author and do not necessarily reflect the
official policy or position of NKS or any of its employees.



This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 15:10:34 EDT