Networks

Ask something
Discussion Locked

NW3.12 vol has excessive vrepair errors

0 helpful
16 replies

by alphaa10

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

NW3.12 vol gets excessive vrepair errors

NW3.12 vol (soon to go to 4.11) must be
vrepaired every day. The thing begs to be
put out of its misery, and I plan to do
that with a conversion over the wire to a
new 4.11 server.

But before that high-wire act, I need to
find out where the vrepair errors are
coming from. A typical day will result in
between 15,000 to 20,000 (you read that
correctly) errors. And, yes, the poor
little vol is backed up religiously every
evening.

My vrepair settings (as I can remember them)
are
1. write all changes to disk immediately
2. no pause between errors
3. keep no error log

16 replies

by kevin anderson

My guess would be a dying hard drive (unless you've changed controller card drivers). You could verify that (more or less) by checking hotfix blocks, which should read 5 or less (0 is perfect).

In any case, I would certainly accelerate the move to 4.11 (or better yet, 5).

It sounds like you know what you're doing in front of a netware box (it's getting rare these days), but make sure you vrepair until you get 0 errors (could be 10 times or more).

If you are seeing hotfix blocks increase, then I would stop doing the vrepair and plan on the migration immediately (it accelerates rapidly).

One other thought. Can you put a second hard drive into the server, and then mirror them? This would at least offer you some security againsta dead HDD.

And as always, make sure you have the latest OS patches, yadda, yadda, yadda.

Kev.

by shanghai sam

Kev,
I should add some information perhaps clarifying the existence of the
"errors" on the hard drive. This NW3.12 sysvol already has been copied
lock, stock (and error?), via PowerQuest's Server Magic 3.0, from an
old 1gb SCSI (circa 1996) to a new Western Digital 4 gb SCSI,
Enterprise class, and run by an Adaptec 2940UW controller.

Because we had anticipated exactly the dangers you describe, this
transfer was made as soon as possible. The move to the new WD SCSI
is still scheduled to be followed by installation of a mirrored drive-- just
as soon as the sysvol can be stabilized. (Had we the money, we would
do a RAID 5). The new WD SCSI 4gb has the standard 3-5 percent
hotfix area set up, already, and all basic patches have been applied
(we were Y2k-ready).

So, that leaves us with problems from a shaky sysvol on a new drive--
puzzling because even if the original, flawed image had been
transferred faithfully by Server Magic (a perfect image of an imperfect
drive), that cannot explain why vrepair could not fix it, once and for all.

We have a condition which shows the following characteristics--
(1) errors accumulate over the interval between vrepairs (2) errors are
"fixed" by each vrepair, at leastin that the utility shows a zero-error
condition when finished (and subsequent test runs of vrepair show
absolutely no change from that zero baseline) (3) error accumulation
appears weakly-correlated with usage/traffic, but averaging 16k-25kerrors per day (4) we have no hardware/equipment hypothesis,
because the only original component left is the mainboard (data
cables have been replaced).

by dcassat

This goes back a ways. Make sure that you have tried the latest version of Vrepair. I know there was a version that would trash files on the hard drive...

Good Luck.

by alphaa10

Thanks, dcassat
According to my reading, that carnivorous vrepair version
accompanied the transition between 4.11 and 5.x, and was a
major reason we have stayed with the original-issue 3.12 vrepair
(1992 or 1993), at least until we are ready to launch into 4.11.

by bob the builder

Have you made a back-up of your "Data"... If so and you are planning to migrate then trash your 4.11 and upgrade to 5.x and copy your data over. This may sound simple but you must realise that the new NOS will give you the most comfort.

by alphaa10

Thanks, Bob
I would love to go with 5.x instead of fertzing around with 4.11,
but the container of which we are a part has standardized for
the time being on 4.11. Not a bad base, but admittedly not as
smooth as 5.x.

by estebandelatorre

AS you can read at novell web site, you may should run many time vrepair in order to have a volumen fixed.
Run the first time whit your current options
when the first finished, start a second one but with a pause between each error, write down each in order to verify if vrepair can or cannot verify the errors.
If none of them could be fixed, I think that you may have:
a.- a HD hardware problem
b.- a Faulty IDE or SCSI controller
c.- A BAD loaded driver
d.- A bad operator that is powering(reseting) down the server afer every volume is correctly dismounted. Many server get hanged while performing a DOWN command.

by alphaa10

Thanks, EDT
We customarily run vrepair until we get a zero error reading, which is
usually on the second run. On only two occasions can I remember
getting a zero error reading the first vrepair run, and my jaw dropped
with astonishment when I witnessed the rare event.

Of course, I was guilty of wishful thinking, as well. Anticipating good
times had arrived at last with the hard disk, imagine my astonishment when the next day, the usual 15k-30k error count returned.

by techdictator

Do you have the option set to purge deleted files? Otherwise there is something wrong with that box and you should do an across the wire migration and send that box in for repairs. Don't wait on it or you may be looking for a new job.

by alphaa10

Thanks, TechDictator,
Yes, the vrepair is always set to purge deleted files.
And yes, we do plan an across-the-wire migration, since it is the most
direct and in this situation, as you point out, there is no future in anything
else.

by techdictator

My point is if you have it configured to purge deleted files, each file is considered to be an error. So the count would be as high as files are deleted/overwritten. If the hotfix area isn't being used for data then you're fine.

by s_pryce

What name spaces are loaded for the vol ?
Number of things could cause that sort of errors.
Not loading the name space module for vrepair and you get 1 error for each file.
Loading the wrong name spaces and 1 for each error.
If thats not it, look at the clients shells in relation to the loaded name spaces. I've seen clients create 1 error for each file they touch.
Anyway what are you running ?

This discussion is closed to new replies.