Re: installation causes power cycle?
- From: "Brian K. White" <brian@xxxxxxxxx>
- Date: Thu, 27 Dec 2007 12:42:18 -0500
----- Original Message -----
From: "RC" <rcaddy@xxxxxxxxxxx>
Newsgroups: comp.unix.sco.misc
To: <distro@xxxxxxx>
Sent: Thursday, December 27, 2007 4:57 AM
Subject: installation causes power cycle?
I purchased a new Intel DG33FBC MB with an Intel E2200 Dual Core
processor and 2 GB memory. I have used a 430watt power supply and a
680 watt power supply. I am installing SCO Openserver 5.0.7 on a
250GB WD hard drive with a SCSI cd-rom, set at id 5 using and AHA-2940
controller. I can get the install cd to boot and I will type in my
bootstring and the system will appear to start the install and then
power off and then back on again. After powering back up it will
typically recycle the power at various stages of booting the cd. I
have actually let it sit at the boot prompt for over an hour to see if
it would reboot, but it did not. Any ideas would be appreciated.
Start with running memtest86 for an hour or preferrably over night.
Then see if anything else reboots it, like the same linux cd you used to get
memtest.
If a linux live cd doesn't cause it to reboot, then:
Perhaps you haven't consulted sco hardware compatibility list?
Granted theres hardly any selection there, but, when you uase _any_ hardware
that hasn't been vetted by someone, be it sco themselves or a hardware
vendor or system builer who builds, tests, and warrantees sco boxes (Seneca
Data is one such), then you takes your chances simple as that. Usually most
things work. But, untill someone proves a particular item, or combo, it
could be incompatible.
If anything else does make it reboot, then, and this is purely hardware
diagnostic 101, not anything sco specific,:
Start with unplugging everything possible from the motherboard but ram,
video, keyboard, and a plain ide cdrom, remove all cards like the 2940 (if
it's built-in, just don't plug any devices in, thus the ide cd). Same for
external cables, unplug and network, serial, printer, scsi... Try to stress
the system using a live-cd. If it still fails under that minimal, stripped
down config, and you can't spot any bios options that help when you change
them, then you can start swapping out those few remaining pieces with
others, find other ram, find another cpu, find another motherboard, swap
them out one at a time until it works, or return it all as defective and
start over.
If it does run ok under that stripped config, then start connecting parts
back on one at a time until it fails again.
Basically you have to conduct a systematic process of elimination to
identify either a defective part, a part that merely SCO doesn't like, or
some combination of parts that might be ok individually or in combination
with other parts, but have some bad interaction together, or merely have a
bad interaction together with each other and SCO.
It's extremely tedious and time consuming unless you luck out and it was bad
ram and the memtest catches it.
That is the most common so it's worth trying, but for the rest, grocery
baggers' time is worth more than spending your life doing that. This is why,
at least for SCO, you buy from some vendor who has already done all that
testing for you and who has the staff and materials to do it so thoroughly
and well that they will warranty the result. Or if you still want to buy any
old random stuff and hope it works, then you must accept the percentage of
failures and just eat them. Usually unless the hardware is actually bad it
will work ok under linux or windows, so you just be prepared to use the
stuff for some other job that needs a linux box and try again with some
other collection of hardware to make the sco box. Probably you won't be
unlucky like that 2 or 3 or 7 times in a row...probably.
I would also look in the bios to disable anything the looks especially new
and unnecessarily fancy, like spread spectrum or speed-step on the cpu or
the pci bus or the ram bus, disable anything that looks like a powersaving
feature anywhere so that the machine runs more statically. Also try
disabling acpi both in the bios if possible, and at the sco boot: prompt,
(the incantations can be googled, I don't know them off hand but I've
googled them ok several times when needed)
Luckily it's probably not some arcane hardware combo, but plain bad ram,
because if it was incompatible hardware it usually wouldn't be intermittent
and random like that. Thats usually bad ram or weak power or extreme age or
heat.
You've already eliminated age, power, and presumably heat. A bad hardware
item or combo will usually have a consistent form of failure, always failing
in the same way and at the same time or under the same set of contitions.
By "set of conditions", I mean this: I had a completely reproduceable and
demonstrable thing once where a known-good 506 or 507 install, before mp1,
on a certain set of hardware, would mostly run fine but would _always_ fail
when trying to uncompress a particular VOL file in a particular update.
Behind the scenes the unpack would produce junk, as if it were a bad file,
but the file was really good.
It turned out to be hyperthreading in the cpu. Disable hyperthreading in the
bios and the custom process always succeeds unpacking the exact same file.
Later sco had an update that fixed that, That kind of software/hardware
interaction could surely have also produced any number random-seeming
failures, could have just as easily caused fs corruption or anything else,
but luckily in that case the failure only happend during a pretty benign
operation (that I know of!) and I spotted the pattern that was there.
That's what I mean by consistent might mean "under the same set of
conditions" not necessarily "at the same time". That box might, did in fact,
run fine for days with all manner of software, but would _always_ fail at
that particular task, which could come at any old time, whenever I happened
to try to do it, (unless hyperthreading was disabled or mp1 was installed)
It was consistent across an entire line of boxes that all had the same
collection of hardware for about a year, not one particular box.
Happy hunting.
Brian K. White brian@xxxxxxxxx http://www.myspace.com/KEYofR
+++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++.
filePro BBx Linux SCO FreeBSD #callahans Satriani Filk!
.
- References:
- installation causes power cycle?
- From: RC
- installation causes power cycle?
- Prev by Date: Re: Very specific question
- Next by Date: Re: SCOX Delisted On NASDAQ (Yahoo Busted Link)
- Previous by thread: installation causes power cycle?
- Next by thread: SCOX Delisted On NASDAQ
- Index(es):
Relevant Pages
|
