[gentoo-amd64] What went wrong

[gentoo-amd64] What went wrong

[Peter Humphrey]

Well, there's a tale and no mistake.

Early December, it was, when I started getting refusal to boot. Even the 
POST wouldn't run until I took out half the RAM. Then, while switching 
sticks about to find where the problem was, in my clumsiness I managed to 
break one of the DIMM slots. Looked like I had to change the motherboard, 
but I wanted advice from the system builder, Armari, so I packed the box up 
and sent it off. The news was not good, so as the system was three years 
old by then, I decided to cut my losses and have the motherboard replaced.

Oh, and by the way, said Darryll, did I know that one of my SATA disks was 
about to fail? I didn't, no. Just disconnect it, I said, and I'll see what 
I can get off it when it comes home. Eight months old - I ask you!

The box came back on the 23rd after quite a wait for SuperMicro to get a 
board out to Armari. I spent spare time over Christmas trying to get a 
system working, and I started an mdraid thread here on 28 December. Thanks 
for the suggestions, those who tried to help.

Problems continued to haunt me. I couldn't boot a floppy with Smart Boot 
Manager on it, nor any other for that matter; I couldn't find a way to 
mount the newly rebuilt md partitions other than from the installation 
disk; and I kept getting mysterious oddities all over the place.

Eventually the penny dropped, and I disconnected the allegedly imminently 
failing SATA disk. I'd already disabled it in the BIOS, but that left the 
electrical connections in place, of course, and I've a nasty feeling that 
stray pulses were getting out of the disk and into places they shouldn't. So 
I pulled both plugs: power and signal, since when the system has seemed 
more stable.

So, one disk down and no RAID. No problem - just delete the type-fd 
partitions off the remaining disk, create new Reiser file-systems and 
restore the backup; then worry about what to do about the possibility of 
the disk failure's having damaged the configuration. Well, as we all build 
from source hereabouts, and the source files are protected with checksums, 
all I had to do, I reasoned, was to emerge -e system and then world.

Heigh-ho. After going through etc-update carefully and making sure I hadn't 
thrown out any of my settings, I now find a few little quirks. I really 
don't want to have install again from scratch, as distinct from recovering 
backups (my wife is patient - of course - but I'm sure she doesn't quite 
understand how running a computer can possibly take up so much time), but I 
can't be confident in the system as it is. Kdm won't start, vmware-server 
locks the machine solid, and one or two other things aren't quite right, 
such as Smart Boot Manager not booting. To cap it all, today both the 
on-board Ethernet ports stopped working and I had to slot in a spare PCI 
card.

We do do all this sort of thing for fun, don't we?

-- 
Rgds
Peter
-- 
gentoo-amd64@gentoo.org mailing list

[gentoo-amd64] Re: What went wrong

[Duncan]

Peter Humphrey <prh@gotadsl.co.uk> posted
200701161712.44900.prh@gotadsl.co.uk, excerpted below, on  Tue, 16 Jan
2007 17:12:44 +0000:

> To cap it all, today both the on-board Ethernet ports stopped working and
> I had to slot in a spare PCI card.

Are you sure your power to wherever you plugin is good?  What about the
onboard power supply?  Is it heavy duty enough to run the RAID and the
system and...?  It's UL (or equivalent in your area) listed for the rated
wattage, right?  What about handling?  You take appropriate static
elimination steps during handling, correct?

As you can already tell from the above, that string of bad luck, with all
that stuff failing, just doesn't seem normal to me.  I once lived on a
mission 50 miles from town, with /bad/ power.  This was back in the 80s,
but we had things like the Commodore64, and the one guy's string of luck
with that thing sure sounds like the historic equivalent of what you just
said.  If it wasn't one thing going wrong, it was another.  He'd have the
thing going a couple days and pow, it'd be out for service for a couple
weeks... or months... depending on whether they had to back-order whatever
it was.

At the same time, the phones out there were failed more than they were
working.  A lot of the time the things would constantly buzz/low-ring.  We
discovered the fastest way to get service was to keep them plugged in
anyway, as that disrupted service further up the line and they'd send a
repair truck out.  We ended up with one of those styrofoam coolers,
stuffed with pillows to deaden the sound further, the buzzing phone in the
center, with the lid on and the thing taped shut.

I went away to school after a year or so, only visiting, but my folks
stayed there for eight years.  By the time they left, there was even a
paved road going by about a mile away, and they'd upgraded both power and
phone so they actually worked as they were designed to work.

So anyway, every time I read about the problems of rural broadband, I
think of that place.  If they can get broadband there, pretty much
everywhere else should have already gotten it.

Back to your situation, however.  It's not electrical spikes from either
the power or whatever broadband you have killing everything, right?  Your
story just sounded too familiar.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What went wrong

[Peter Humphrey]

On Tuesday 16 January 2007 21:42, Duncan wrote:

> Back to your situation, however.  It's not electrical spikes from either
> the power or whatever broadband you have killing everything, right?  Your
> story just sounded too familiar.

In this country the power supply system is quite civilised. To illustrate: I 
remember quite a heated discussion involving an American who was adamant 
that no supplier of surge suppressors could possibly offer a replacement 
guarantee on any equipment damaged by spikes when connected to its 
suppressors. Not and stay in business, that is. Over here, though, I have 
such a guarantee certificate in the filing system which relates to the 
surge suppressor I use to supply my installation, and such guarantees are 
common. It undertakes to replace any equipment damaged by a spike, up to a 
value of £20,000. What I'm trying to say is that I can rule out waywardness 
of the public electricity supply. Oh, and the phone line also passes 
through a section of the suppressor and I'm confident that I'm not being 
spiked that way either. And we haven't had lighting hereabouts for quite a 
while anyway. Drizzle, yes, but not lightning.

No, the problems are local. That disk has done some damage, I'm fairly sure 
but with no real evidence, and I probably haven't helped either. Yes, I do 
use an earthing strap (or sometimes I take a short-cut and keep one hand in 
contact with the framework), but somebody suggested the other day that my 
floppy disk drive's signal cable might have been connected the wrong way 
round (what price keys, eh?), so I tried it the other way, thinking that 
the symptoms could fit. That didn't help, so I put it back again and since 
then the Ethernet ports have been dead. Defensive design? What's that?

Sometimes, when nothing works, it's hard to find a way in so as to fix it.

-- 
Rgds
Peter

-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What went wrong

[Peter Humphrey]

On Tuesday 16 January 2007 21:42, Duncan wrote:

> Are you sure your power to wherever you plugin is good?  What about the
> onboard power supply?  Is it heavy duty enough to run the RAID and the
> system and...?  It's UL (or equivalent in your area) listed for the rated
> wattage, right?

What's UL?

-- 
Rgds
Peter
-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What went wrong

[Ryan Sims]

On 1/16/07, Peter Humphrey <prh@gotadsl.co.uk> wrote:
> On Tuesday 16 January 2007 21:42, Duncan wrote:
>
> > Are you sure your power to wherever you plugin is good?  What about the
> > onboard power supply?  Is it heavy duty enough to run the RAID and the
> > system and...?  It's UL (or equivalent in your area) listed for the rated
> > wattage, right?
>
> What's UL?

Underwriters Laboritories, iirc.  They basically make sure things work
like they're supposed to, i.e. without killing anyone or destroying
other equipment.  I think the equivalent across the pond is CE (I
think).

http://en.wikipedia.org/wiki/Underwriters_Laboratories

-- 
Ryan W Sims
-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What went wrong

[David Fellows]

On Wed, 17 Jan 2007 00:04:43 +0000 
Peter Humphrey wrote -
> On Tuesday 16 January 2007 21:42, Duncan wrote:
> 
> > Are you sure your power to wherever you plugin is good?  What about the
> > onboard power supply?  Is it heavy duty enough to run the RAID and the
> > system and...?  It's UL (or equivalent in your area) listed for the rated
> > wattage, right?
> 
> What's UL?

Underwriters Laboratories.  They test and certify products for electrical 
and fire safety and adherence to relevant safety standards and codes. 
As far as I know they do not comment on performance.  Ie a UL listed power 
supply shouldn't electrocute you nor spontaneously combust, but there are no
statements about how well it supplies regulated power to your motherboard.

 
Dave F
-- 
gentoo-amd64@gentoo.org mailing list

[gentoo-amd64] Re: What went wrong

[Duncan]

David Fellows <fellows@unb.ca> posted
200701170045.l0H0joJQ024617@mailserv.unb.ca, excerpted below, on  Tue, 16
Jan 2007 20:45:49 -0400:

> Underwriters Laboratories.  They test and certify products for electrical
> and fire safety and adherence to relevant safety standards and codes. As
> far as I know they do not comment on performance.  Ie a UL listed power
> supply shouldn't electrocute you nor spontaneously combust, but there are
> no statements about how well it supplies regulated power to your
> motherboard.

Except that they do certify the supply as providing the rated wattage and
maintaining voltages under load.  The reason is that if a power supply
won't meet its ratings, only providing a sustained 250 watt when it's
rated 450 watt, if it's actually hooked to a load demanding 450 watt, it's
overloaded seriously enough to have a good chance of shorting out. 
Anybody like exploding power supplies?  What about when you are sleeping
or at work, and left the computer running?  That's a good way to lose a
house to fire, or short out hot to the exposed metal chassis for some
two-year-old kid to come across and fry themselves.  /That's/ why UL has
an interest in and rates computer power supplies, among other things.

The worst part is, a lot of cheap cases come with power supplies with
/incredibly/ inflated ratings.  That 450 watt rating but only providing
250 watt is actually not uncommon at all, in such things.  However, when
one examines said power supplies, they tend to have no recognized
certifications of any sort.  They stick some random but good sounding
rating on the side, but the things are crap and that's pretty much exactly
what it is, a random but good sounding rating.

If one always ensures they get a UL (or whatever local equivalent)
certified power supply, it certifies at least two things.

1) The power supply provides the rated voltages under the specified loads
(wattage/amperage).

2) Should some serious over-voltage or over-amperage (including
load side dead-shorts to ground) occur, the power supply will NOT fail
catastrophically. That is, it will NOT explode, shoot fire, or otherwise
become a hazard to human health and safety.  Many will blow fuses or
better yet, trigger electronic shutoffs of the shorted out leg, such that
resetting the shutoff or replacing the fuse will get you back in business,
but at worst, the device must do no more than "brick", that is, fail with
little more than the pop of critical components deliberately designed in
at the failure points, while continuing to protect human health and safety.

It should be noted what the UL does NOT do, however.  While they certify
sane operational ranges withing rating, and non-catastrophic failure
modes, they do NOT necessarily certify voltage rise times and the like
specifically suitable to computerized electronics.  It's theoretically
possible to get a UL certified power supply that simply isn't suitable to
power a computer, even tho it meets safety and specific rating
certification requirements.  However, in practice, the companies that go
to the trouble and testing expense of getting UL certified, aren't the
type of fly-by-nights that simply don't care, or they'd not have gone to
the trouble and expense in the first place.  By the time they get their
products UL certified, they are companies that have invested significant
resources into their reputations, and have no intention of blowing that,
having gone to all the expense to design and certify their product, just
to save 10 cents on a component that while not dangerous, won't provide
suitably stable computer operating power.  Thus, in practice, UL listed
means rather more than the actual certifications in the test itself, and a
UL listed power supply is with little doubt going to be a dependable
power supply, as well as physically safe.

OTOH, trust your computer to an uncertified power supply and you are very
literally gambling not only the computer's safety but your own.   Google
exploding power supplies some time, should you doubt.

Of course, that's pretty much what you said anyway, except that you didn't
note the practical effect of UL certification being a mark of quality,
because certification has by definition and process already separated out
the guys who don't care about reputation or quality and are simply out to
make some fast money, human safety or no human safety.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What went wrong

[Peter Humphrey]

On Wednesday 17 January 2007 00:45, David Fellows wrote:
> On Wed, 17 Jan 2007 00:04:43 +0000
> Peter Humphrey wrote -
>
> > What's UL?
>
> Underwriters Laboratories.  They test and certify products for electrical
> and fire safety and adherence to relevant safety standards and codes.
> As far as I know they do not comment on performance.  Ie a UL listed
> power supply shouldn't electrocute you nor spontaneously combust, but
> there are no statements about how well it supplies regulated power to
> your motherboard.

Thanks.

I don't know of an equivalent to UL over here, but it's a SuperMicro case 
round a system built by Armari, a reputable workstation builder, so I'm not 
worried about the quality of the PSU.

-- 
Rgds
Peter
-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What went wrong

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 820 bytes --]

On Wed, 17 Jan 2007 00:00:45 +0000, Peter Humphrey wrote:

> > Back to your situation, however.  It's not electrical spikes from
> > either the power or whatever broadband you have killing everything,
> > right?  Your story just sounded too familiar.  
> 
> In this country the power supply system is quite civilised.

Maybe, but not as good as it used to be. Brownouts are far too
commonplace nowadays. After moving house a few years ago, less than a
mile, I experienced random hardware failures; hard disks, motherboard,
even two 22" monitors (fortunately still in warranty). I suspected bad
power and, touch wood, everything has been reliable since fitting a UPS.

I now have to create my own failures in software :)


-- 
Neil Bothwick

Windows Multitasking - screwing up several things at once

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-amd64] Re: What went wrong

[Peter Humphrey]

On Wednesday 17 January 2007 09:47, Neil Bothwick wrote:
> On Wed, 17 Jan 2007 00:00:45 +0000, Peter Humphrey wrote:
> >
> > In this country the power supply system is quite civilised.
>
> Maybe, but not as good as it used to be. Brownouts are far too
> commonplace nowadays. After moving house a few years ago, less than a
> mile, I experienced random hardware failures; hard disks, motherboard,
> even two 22" monitors (fortunately still in warranty). I suspected bad
> power and, touch wood, everything has been reliable since fitting a UPS.

That's the first I've heard of it. Maybe I should look into a UPS for 
myself.

> I now have to create my own failures in software :)

Oh, that's the easiest part!

-- 
Rgds
Peter
-- 
gentoo-amd64@gentoo.org mailing list