SCSI disk problem

Thread View: comp.sys.sgi.hardware

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by Rainer Kleinrensing Sat, 02 Jan 1993 15:07

#45

Author: Rainer Kleinrensing
Date: Sat, 02 Jan 1993 15:07

45 lines
2074 bytes

Hello,
After christmas we found that our 4D-220VGX had crashed in a mysterious way:
many programs of the OS could not be found any more, we had to reload 
the root partition from the last backup. In SYSLOG there were the following
messages:
Dec 27 03:13:19 gt unix: WARNING: dma_map: address not word aligned
Dec 27 03:13:19 gt unix: sc0,1,0: cmd=0x28 disconnected on non-word boundary (ad
dr=C046AE4F, 0x0 left)
Dec 27 03:13:21 gt unix: .  Resetting SCSI bus
Dec 27 03:13:21 gt unix: dks0d3s0 (/usr): retrying request
Dec 27 03:13:21 gt unix: dks0d1s6 (/usr/amandus): retrying request

(repeated several times) and

Dec 27 05:02:12 gt unix: dks0d1s0 (/): Directory 13 is corrupted (4d)
Dec 27 05:02:13 gt unix: dks0d1s0 (/): inum: 19 illegal number of indirect exten
ts
Dec 27 05:02:13 gt unix: 
Dec 27 05:02:13 gt unix: dks0d1s0 (/): inum: 20 illegal number of indirect exten
ts

(also with some other numbers instead of 19,20)

After restoring the backup everything worked just fine. However, a few
minutes ago I saw 
Jan  1 05:00:15 gt unix: dks0d1s6 (/usr/amandus): inum: 1724 illegal number of i
ndirect extents
Jan  1 05:00:15 gt unix: 
Jan  1 05:00:16 gt unix: dks0d1s6 (/usr/amandus): inum: 1725 illegal number of i
ndirect extents

in SYSLOG. Is this just a sign of the disk going bad (I admit it's no SGI disk,
but has workes flawlessly for at least 3 years now) or did we
inadvertently make some configuration or software error ?

Thanks very much for any hints,



-- 

   Rainer Kleinrensing,    rainer@indigo.mathematik.uni-freiburg.de
   Institute for Applied Mathematics
   Freiburg University

Re: SCSI disk problem

#47

Author: Dave Olson
Date: Sun, 03 Jan 1993 21:28

48 lines
2910 bytes

In <1993Jan2.150746.6429@sun1.ruf.uni-freiburg.de> kleinren@sun1.ruf.uni-freiburg.de (Rainer Kleinrensing) writes:
| After christmas we found that our 4D-220VGX had crashed in a mysterious way:
| many programs of the OS could not be found any more, we had to reload 
| the root partition from the last backup. In SYSLOG there were the following
| messages:
| Dec 27 03:13:19 gt unix: WARNING: dma_map: address not word aligned
| Dec 27 03:13:19 gt unix: sc0,1,0: cmd=0x28 disconnected on non-word boundary (ad
| dr=C046AE4F, 0x0 left)
| Dec 27 03:13:21 gt unix: .  Resetting SCSI bus

In the case of disks, this is almost always a SCSI bus error of some
sort (bad cable, loose connector, missing or extra terminators), since
they almost always disconnect on a block boundary.

SGI's DMA hardware requires word (low 2 bits of address 0) aligned
addresses to work; this message tells you that the disconnect was
not on a word boundary.  The 0 left part is a bug in my code; it *always*
prints 0 in some releases; I think I fixed it for 405, but it might
have been 405F.

| Dec 27 05:02:12 gt unix: dks0d1s0 (/): Directory 13 is corrupted (4d)
| Dec 27 05:02:13 gt unix: dks0d1s0 (/): inum: 19 illegal number of indirect exten
| ts
| Dec 27 05:02:13 gt unix: 
| Dec 27 05:02:13 gt unix: dks0d1s0 (/): inum: 20 illegal number of indirect exten

This could be the result of cabling problems also, if bad data got written
to disk.  It could also be a firmware problem with the drive.

| After restoring the backup everything worked just fine. However, a few
| minutes ago I saw 
| Jan  1 05:00:15 gt unix: dks0d1s6 (/usr/amandus): inum: 1724 illegal number of i
| ndirect extents
| Jan  1 05:00:15 gt unix: 
| Jan  1 05:00:16 gt unix: dks0d1s6 (/usr/amandus): inum: 1725 illegal number of i
| ndirect extents
| 
| in SYSLOG. Is this just a sign of the disk going bad (I admit it's no SGI disk,
| but has workes flawlessly for at least 3 years now) or did we
| inadvertently make some configuration or software error ?

Did you do either an mkfs or an fsck before doing the restore?  If so,
either the drive is failing, or something has recently changed about
the physical setup on the machine, that is affecting the SCSI bus.
--
Let no one tell me that silence gives consent,  |   Dave Olson
because whoever is silent dissents.             |   Silicon Graphics, Inc.
    Maria Isabel Barreno                        |   olson@sgi.com

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads