Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!samsung!uunet!mcsun!ukc!stl!crosfield!adf
From: adf@crosfield.co.uk (alex france)
Newsgroups: comp.unix.wizards
Subject: Re: CAN NOT READ BLK xxxxx is fsck
Message-ID: <10387@suns7.crosfield.co.uk>
Date: 24 Jun 91 08:42:50 GMT
References: <27262@adm.brl.mil>
Reply-To: adf@crosfield.co.uk (alex france)
Followup-To: comp.unix.wizards
Organization: Crosfield Electronics, Hemel Hempstead, United Kingdom.
Lines: 45

In article <27262@adm.brl.mil> ESANCHEZ@udlapvms.pue.udlap.mx writes:
>
>>When doing  "fsck /dev/rusr1", I get three messages like the following:
>>   CAN NOT READ BLK 31993
>>Two occur in Phase 1, then in Phase 5 I get another CAN NOT READ message,
>>followed by a message saying
>>   4437 BLK(S) MISSING
>>   BAD FREE LIST
>>   SALVAGE? y <---------
>> 
>>No matter how much I try running fsck, I keep getting these errors.  Does
>>anyone out there have suggestions how to fix this?
>> 
>	Just answer "yes" to salvage question and that's all. At least,
>I have 21 Sparcs and sometimes I got that messages ( When the light
>goes down and the machines stops without the "shutdown") and I answer yes
> ........

Seems to me like the man has already tried this, with no success. I have
found, in the past, that it is possible to get real hard disk errors
on SCSI disks which leave unreadable (but writeable) disk sectors. The
controller thinks it wrote the block OK, but 5 minutes later you
cannot read it. So, endless typing of 'Y' to fsck does not improve
matters. Also, some device drivers do not automatically perform 'grown
defect mapping' on SCSI disks.

Given the large number of blocks it thinks are missing from
the free list, my guess would be that you are losing an inode block,
although you don't say you are getting any messages about missing
files.

Anyway, and unfortunately, the only way to resurrect such a problem is
to find something that will scan your disk and add grown defects (this
remaps unreadable sectors onto spare sectors in the same cylinder).
Usually this is found within disk diagnostics or formatter programs,
but I don't know where this might be in SCO Unix (anyone else?). You
may find that you have to do a full reformat and surface scan on the
disk to clear the problem - a long and painfull business if you have
to back everything up first.

Hope this helps.

Alex France, Crosfield Electronics, | Phone: +44 442 230000 xt.3541
Hemel Hempstead, HP2 7RH, England.  | Fax:   +44 442 232301
adf@crosfield.co.uk or mcsun!cel!adf@uunet.uu.net