hurf@batcomputer.tn.cornell.edu (Hurf Sheldon) (08/17/87)
Our uVaxII - Ba123/Ultrix1.2 system with decqna,9meg
3 dhv, 1 dz, 1 qdss, 1 rqdx3/2 rd53
1 tqk50 and one SI/Sigma RQD11-EC/mx4380;
We have had a disk problem after a power outage - Our
SI/Sigma RQD11-EC/Maxtor 4380 failed fsck on autoreboot.
Running fsck manually got "Unable to read blk#nnnnnnn where
nnnnnnn was any alternate superblock in any partition except
the 'b' partition. The 'b' partition could find superblock 32
but found all the inodes to be in error. There were some processes
running at the time and the 'a' partition is used as swap space.
The disk was partitioned according to the following table using chpt(8)
and has worked without error for several months. Both rqdx3/rd53 systems
we have survived without error with swap on the 'a' partition on the
second rd53 in both cases. (One on this uVaxII one on another uVaxII)
#
mx4380|MX4380|Winchester Maxtor4380-Sigma SDC-RQD11-EC:\
:ty=winchester:ns#34:nt#15:nc#1224:\
:pa#31768:ba#4096:fa#1024:\
:pb#147727:bb#4096:fb#1024:\
:pc#622676:bc#8192:fc#1024:\
:pd#147727:bd#4096:fd#1024:\
:pe#147727:be#4096:fe#1024:\
:pf#147727:bf#4096:ff#1024:\
:pg#295454:bg#8192:fg#1024:\
:ph#590908:bh#4096:fh#1024:
Running the nondestructive diagnostic on the controller (Wombat)
found no read errors or ROM/CACHE errors.
There are no 'write' errors logged in /usr/adm/messages or syslog.
The thing that concerns me is that the disk appears to have been
completely trashed as a result of the power going down. We can survive
this one ok but would like to prevent this in the future.
Would also like to know what happened before we restore.
--
Hurf Sheldon Network: hurf@ionvax.tn.cornell.edu
Lab of Plasma Studies Bitnet: hurf@CRNLION
369 Upson Hall, Cornell University, Ithaca, N.Y. 14853 ph:607 255 7267
I sold my Elan, got a job in science; Now, no one takes me seriously.hurf@batcomputer.UUCP (08/18/87)
In article <2037@batcomputer.tn.cornell.edu> hurf@tcgould.tn.cornell.edu (Hurf Sheldon) writes: > > Our uVaxII - Ba123/Ultrix1.2 system with decqna,9meg > > We have had a disk problem after a power outage - Our > SI/Sigma RQD11-EC/Maxtor 4380 failed fsck on autoreboot. > Running fsck manually got "Unable to read blk#nnnnnnn where > nnnnnnn was any alternate superblock in any partition except > the 'b' partition. The 'b' partition could find superblock 32 > but found all the inodes to be in error. There were some processes > running at the time and the 'a' partition is used as swap space. It seems that the only pointers to the disk partition structure were in the 'a' partition which were written over by swapping. After the crash the driver couldn't find the partition table and decided based on what I am not sure to use ra81 default partition tables. Restoring the partition table with chpt(8) on the 'a' partition solved the problem and we were able to run fsck(8) successfully. > > The disk was partitioned according to the following table using chpt(8) > and has worked without error for several months. Both rqdx3/rd53 systems > we have survived without error with swap on the 'a' partition on the > second rd53 in both cases. (One on this uVaxII one on another uVaxII) Question: Why do the rd53's work ok and why does fsck not get lost on a normal shutdown? Does synching the disk write the partition table somewhere? > > Observation: It is not a grand idea to swap to the 'a' partition despite the (small) performance gain. Question: How would one most appropriately set up for a new disk description? By altering the driver? But then you need sources. >-- -- Hurf Sheldon Network: hurf@ionvax.tn.cornell.edu Lab of Plasma Studies Bitnet: hurf@CRNLION 369 Upson Hall, Cornell University, Ithaca, N.Y. 14853 ph:607 255 7267 I sold my Elan, got a job in science; Now, no one takes me seriously.