[Tfug] Check file system and restore array

[Bexley Hall]

Hi Timothy,

On 2/9/2013 5:52 PM, Timothy D. Lenz wrote:
> Grub as always given an error about the floppy at boot. It does a drive
> seek, then says something like fd0 error, then boots from the hard

Is it an *error* or an informative message?  And, you are actually
hearing/seeing the drive spin up and seek?

> drives. The cpu is an AMD Athlon 64x2 4200+ dual core. Motherboard is

So, reasonably new.

> Asus M2N-E. And yes it runs 24x7 and is on a UPS. I built 2 of these,
> one for me and one for my dad, but then he decided to just use hulu. The
> one I built for him just has 2 WD 500Gb, drives in raid 1 and basicly
> the same cpu, only a 5000. How I ended up building that second system is
> another long story not really relevent cept it is the reason I will
> never buy another asus motherboard. His system was built with 32bit
> linux, mine was built with 64bit. Oh, and I built my system in a 4U rack
> case with very good cooling.
>
> Buying anymore new hardware is not an option. I spent more then I could
> afford already on this problem when it was in the shop. Mixing parts is

I didn't suggest that you throw money at it!  But, if you're not
spending *money*, you have to spend *time*!

> also not much of an option, The board in his system won't boot with ECC
> memory set to ECC. Won't get past post if ECC is turned on in cmos. It

Shirley those aren't the only "swapping" options available to you?!
E.g., you could put his MB in your machine, your drives on his MB,
his drives on your MB, etc.

Your comments suggest that this isn't a *software* problem so you
need to explore different hardware options (or, "restore" everything
and *hope* it doesn't happen again -- without ever understanding
*why* it happened in the first place).

When I bought my first "PC", I bought *two* of them.  *Identical*
configurations.  At that time, PC's were still struggling with
compatibility issues, poor quality clones, flakey software, etc.
I was constantly having to bail out friends/colleagues who'd found
themselves with a "misbehaving" system (usually in the wee hours of
the morning before a dog-and-pony, etc.).

So, I decided it was worth the extra money ($8K at that time) to have
a "second opinion" that I could consult whenever something "wasn't
quite right".  I.e., if both machines exhibited the same problem,
it was most likely a software bug whereas a problem confined to a
single machine suggested something in *that* machine was amiss
(though it could still be software -- but, really easy to compare
the current state of the software between the two machines... much
easier than comparing the functionality of two different pieces of
*hardware*!  :> )

> also doesn't reconize this cpu correctly. They screwed somethings up in
> the newer bios it has.

But you could set *your* drives aside, verify *his* drives still
work in *his* machine, then move them to *your* machine and see if
your machine will work with (ahem) "good" drives.  I.e., if you start
seeing the same sorts of problems with *his* drives, then you can be
suspicious that the problem lies in your machine and not just a
"flakey seagate drive(s)".

> I did get the old power supply back, but I just put it away. I have
> worked on switch mode power supplies many times in VCRs and once parts
> start going bad in them, it's best not to bother with it. I know about

I'm not suggesting you *repair* it.  Rather, it's "evidence" of a
possible failure.  *And*, a hint as to *other* problems that can
lie ahead!

E.g., caps failing in a power supply can result in high (i.e., higher
than tolerable!) ripple voltages on individual supplies (esp if the
"failed" supply is the one that the switcher tracks to regulate all
of the *other* supply voltages!).  With today's lower voltage devices,
it's easy to exceed "absolute maximum ratings" on components.  Or,
stress them by operating them in these atypical conditions.  When
you consider the low core voltages used in modern CPUs and the
proximity (electrically and physically) of these failed caps to the
CPU itself... (ditto for GPU's)

> caps testing ok, even looking ok, no domed tops, swrunk skin etc. It
> takes a $1000 cap checker to properly test them. The high frequency

You can check ESR with an RLC meter (if it is capable of measuring
dissipation factor).  Or, a better/quicker way is to just look at the
anodes with a scope, in situ (i.e., magnitude of ripple *voltage*).

> heats them internally drying them out. Always use 105 degree rated caps
> in switch modes.

It's not the frequency that hurts (though frequency comes into play in
terms of the device's ability to handle a particular current).  Rather,
it is the magnitude of the ripple current -- i.e., power ("heat")
varies as the square of the current.  As switching frequency increases
(all else being -- <cough> -- equal), ripple current *decreases*.

> I think any extra drives I have laying around are IDE. They may not show
> a problem if it's a sata problem. I'd have to dig around and see what I
> have. Would be good to have for testing new kernel builds though
> sometime down the road.

I don't like the common practice of "swap to repair" -- since it
gives no reassurance that the problem has *truly* been repaired
(and isn't just being *masked* by the "new environment").  E.g.,
*how* did the power supply fail?  *Why*?  And, what are the
likely consequences of that failure for the rest of the system?
(repeat for any other "failure")

Fixing the "same problem" twice is a sure indication that it wasn't
*fixed* the first time!  (or second??  :> )

Doesn't cost much (besides time) to move drives around and see if
the problem moves with them!  E.g., rebuild your system with your
Dad's drives and run it that way for a while.  You'll get the *use*
of the system while "exercising" it.

If you have *any* sort of spare drives (even IDE) available,
consider imaging the drive *after* you've built the system
so you can quickly rebuild (restore image) that system later
(even if you move to *other* drives!).  I keep a slew of
external USB drives for just this purpose.  And, a shoebox of
"bare" SATA drives that I use to store the "virgin" images of
each machine that I build (plugging them into a USB SATA
"cradle" when I need to access their contents).  E.g., restoring
an image happens at about 1GB/min compared to the many *hours*
it might have taken to install the OS, applications and then
"tweek" it all for my particular needs!

Clonezilla is your friend.

--don