Help with disquieting error messages and a bad disk block

jensen jensen at slyrf.dkuug.dk
Sun Aug 12 23:08:08 AEST 1990


Synopsis:

I'm seeking enlightenment on the following two problems we are having with
the Sun-3's here:

1. I see esoteric error messages from the kernel that might be
   significant, but I haven't got any idea what they mean.

2. Need to rescue a disk from a bad block, knowing the absolute block
   number.

(1)

We've got two Sun-3's running here, and from time to time they deliver a
short series of disquieting error messages on the console.  These messages
are of an esoteric nature, at least they seem so to me :-) and I don't
know their significance.

We get the following two messages:

     le0: WARNING: if_snd full
     zs0: silo overflow  [or:] zs1: silo overflow

and recently this gem:

     itrunc: /usr/71999 new size = 0, blocks = -16

Any explanation or help with these will be appreciated.  (Maybe I should
add that the Suns are on an Ethernet, with some directories
cross-mounted.)

(2)

Recently one of the Suns began reporting a bad diskblock in the following
manner:

       sd0:  warning, abs. block 37586 has failed 14 times 
       [etc]

And recently escalating to:

       sd0:  block 0x92d2 needs mapping
       sd0:  warning, abs. block 37586 has failed 22 times

I've asked around and received the advice that it would be a good idea to
locate the file (if there is such a file) that makes use of the bad block,
and get that file moved out of the way - to /dev/.bad/<blockno.> or the
like.

My plan of action was to use icheck(8) to find the inode, then use find(1)
to locate the file.  But if I interprete the output from dkinfo(8)
correctly (see below), then abs. block 37586 is positioned in the paging
area.  (The disk is configured in a standard manner, I think, with / on
sd0a and /usr on sd0g.)

After the kernel told me that the block "needed mapping" I did a man(1) -k
on `map' and `mapping', but it didn't turn up anything useful as far as I
could see.

Nothing to do but holler for Technical Support?

------- Output from dkinfo(8):

sd0: Emulex MD21 controller at addr 140000, unit # 0
1544 cylinders 15 heads 50 sectors/track
a: 16500 sectors (22 cyls)
   starting cylinder 0
b: 64500 sectors (86 cyls)
   starting cylinder 22
c: 1158000 sectors (1544 cyls)
   starting cylinder 0
g: 1077000 sectors (1436 cyls)
   starting cylinder 108

------- end of output from dkinfo

Suggestions will be welcomed.

(jensen at dkuug.dk)



More information about the Comp.sys.sun mailing list