Many thanks to
1) "Christopher L. Barnard" <cbar44@tsg.cbot.com>
2) Russ Poffenberger <poffen@San-Jose.ate.slb.com>
3) ibmsuxwg-C.KumarAH0224XXXX
<chenthil@mtcts1.mt.lucent.com>
It all seemed the disk was deteriorating fast, so we
changed it.
Original question:
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
a) I'm getting this error from /var/adm/syslog (SPARC20,
Solaris 2.5.1)
I looked through many newsgroups and former this-list
summaries but found
nothing about it.
WARNING:
/iommu@f,e0000000/sbus@f,e0001000/espdma@f,400000/esp@f,800000/sd@2,0 (sd2):
Error for Command: read Error Level: Fatal
Requested Block: 1642432 Error Block: 1642432
Vendor: MICROP Serial Number: 1936-21
Sense Key: Illegal Request
ASC: 0x24 (invalid field in cdb), ASCQ: 0x0, FRU: 0x0
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
b) format/analyze says the disk's surface is OK. Repairing
the "defective"
block didn't make the WARNING dissapear.
1) "Christopher L. Barnard" <cbar44@tsg.cbot.com>
The format/analyze/repair section of the "format"
command would give a clean
bill of health to a piece of cardboard. It is completely
worthless for
mapping out bad sectors in a disk.
2) Russ Poffenberger <poffen@San-Jose.ate.slb.com>
What size disk is it? An "Illegal Request" usually indicates
an attempt to
access past the size of the disk. This would indicate a
format.dat/partitoning
problem. The problem block is 1642432, so if the disk is
~800Mb or smaller, this
is probably the problem.
3) ibmsuxwg-C.KumarAH0224XXXX
<chenthil@mtcts1.mt.lucent.com>
See how many partitions/Filesystems are on Disk2.If there
are less number of
those & if it's possible for U to back them completely usin a
dd,then do that &
do a complete format & newfs on those partitions on that
disk.Then restore U'r
backup & C if it works.
I add: I failed to mention the disk was actually >2Gb and
worked fine through
several years (two, actually).
* First Sector Last
* Partition Tag Flags Sector Count Sector Mount
Directory
0 2 00 0 4233110 4233109 /pj
2 5 01 0 8467850 8467849
3 0 00 4234740 4233110 8467849 /pk
As you see, the failing sector lies in the first third of the disk.
c) In order to evaluate que quality of our last backups (it
took some time for us
to perceive the error), I would like to know how HARD this
error is.
I mean, did the process which got this error finally read the
RIGHT info off
this block (maybe through some retrying or unknown
principle)?
Or can I safely assume the entire read file is corrupt? In the
latter case,
why is it labeled as just WARNING?
1) "Christopher L. Barnard" <cbar44@tsg.cbot.com>
It isn't "just a warning". WARNING is the most serious flag
on a disk
failure message in your syslog. If you see a WARNING,
you need to act on
it immediately.
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
d) What can I expect from a backup made with this
message?
1) "Christopher L. Barnard" <cbar44@tsg.cbot.com>
Your backups are probably ok, but you really should
replace that disk.
My experience has typically been that my backups are ok,
but that the
data on the disk is not always usable. Your best bet is to
get that disk
replaced as soon as you can and then restore it from your
backup tape.
I add: After your discussion, I would think my backups are
*structurally OK*, but
I don't have much hope the data within the tape is usable. I
mean, if a WARNING is severe
as you said and the disk didn't come to get read OK by the
backup process (action which
triggered the warning), I cannot guarantee the quality of the
reading.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:59 CDT