Subject: From: Date:

RE: [ECS] New beta problem Ingo Pakleppa Mon, 31 May 1999 12:52:29 -0700

---

You are right - a swap file on a separate partition won't fragment, or at
least not as quickly as swap files on the main drives. Depending on how
much other disk activity you have, that may or may not be a problem. The
reason that fragmentation can be bad is that it makes your hard disk seek
back and forth all over the disk - each change of cylinder is expensive in
terms of time. So - if you have a lot of disk activity on your C, D and E
drives, you won't get the benefit of an unfragmented swap file on drive F
because the hard disk will have to go back and forth between C, D and E.
Well, and if the swap file is too large, you'll incur the additional
overhead of Windows managing the swap file - in fact, Windows can get
confused if that happens.

Bottom line: it's as much an art and guesswork as a science about what
really is optimal. My own *guess* is that the best thing is to just let
Windows manage the swap file, and regularly defrag your disks (use the
scheduler from the 95 Plus pack to do that automatically overnight).

Ingo

At 12:56 PM 5/31/99 -0500, you wrote:
>The F drive is a separate partition, not a separate physical drive.  Windows
>does manage the swap file but I have "asked" it to manage it there.  I
>always thought (I am using w98) that the benefit of a separate partition was
>no fragmentation of the swap file?
>
>Mark S
>
>-----Original Message-----
>From: Ingo Pakleppa [mailto:ipakleppa@home.com]
>Sent: Monday, May 31, 1999 12:10 PM
>To: ecs-list@netbloc.com; ecs-list@netbloc.com
>Subject: Re: [ECS] New beta problem
>
>
>Just as a troubleshooting hint: I found that this "invalid page fault" is
>sometimes caused by a stack overflow - too many function calls nested too
>deeply. Or it could be a dangling pointer or the like, causing memory
>corruption.
>
>Other possible causes are:
>
>- drive with the swap file really is full.
>- disk corruption on the swap drive
>- if specifying the swap file size manually, specifying a wrong size (too
>small OR too large could both cause problems under 9x).
>
>By the way, you mention that the F drive is used exclusively as the swap
>drive. If you are using 95 or 98, I would suggest you let Windows manage
>virtual memory automatically - it works more reliably, and will probably
>also be faster if the swap file is wherever Windows puts it (unless the F
>drive really is a separate physical hard disk).
>
>Ingo
>
>At 05:18 AM 5/31/99 -0700, Mark Gilmore wrote:
>>Mark - I suspect that the problem is somehow related
>>to your "swap disk". Can you disable this for a while ?
>>Also, did you run that VR test that I requested ?
>>Thanks.
>>
>>Mark Sekelsky wrote:
>>>
>>> I have awoke to a non-functioning ECS the last two mornings, although
>each
>>> morning was a littel different.
>>>
>>> Sunday morning I had an error dialog that said my "F" disk was full and I
>>> should initiate a disk cleanup.  I have one hard drive and it is
>partitioned
>>> as "C" "E" and "F."  The "F" partition is about 180 megs and is used
>>> exclusively for the Windows swap file.
>>>
>>> Today, Monday, I got the same thing plus a few more errors.  I had a
>dialog
>>> that an illegal error had been caused by VCMD (I might have those letters
>>> mixed up).  Unfortunately, I couldn't get the stack dump.  I also had an
>>> illegal error by ECS.  The dump is:
>>>
>>> caused an invalid page fault in
>>> module MSVCRT.DLL at 0177:78001648.
>>> Registers:
>>> EAX=00e0ba74 CS=0177 EIP=78001648 EFLGS=00010206
>>> EBX=00000000 SS=017f ESP=0172fe1c EBP=0172fe24
>>> ECX=00000019 DS=017f ESI=00e0ba10 FS=2e97
>>> EDX=00000000 ES=017f EDI=00000000 GS=0000
>>> Bytes at CS:EIP:
>>> f3 a5 ff 24 95 28 17 00 78 8b c7 ba 03 00 00 00
>>> Stack dump:
>>> 00000000 00d43ac0 0172ff14 63683880 00000000 00e0ba10 00000064 00000028
>>> 00d43ac0 bff7c79d 00000001 78035108 78035008 00000006 00e3d012 0172fe80
>>>
>>> Becasue of the issues I have had wih the new VR (and that out-of disk
>space
>>> error) it ismy prime suspect. when I rebooted Sunday I made sure that
>>> nothing other than ECS and ACE were running (normally I havea couple of
>>> small utilites running inthe back ground).  I am going to run today
>WITHOUT
>>> VR running and see what happens.  Beyond that I am unsure how to
>>> troubleshoot this one.
>>>
>>> Mark S
>>
>>--
>>Mark Gilmore
>>Omnipotence (ECS Home Automation Software)
>>omnip@usit.net
>>http://www.usit.com/omnip

---