Multiple SCSI Disks on Esix system

heiser at tdw201.ed.ray.com heiser at tdw201.ed.ray.com
Fri Mar 8 02:51:24 AEST 1991


Sorry if this appears twice ... I am re-posting from another site since
my /usr/spool/news directories are pretty badly corrupted.

I recently installed a second scsi disk on my Esix system -- it is a 150mb
Toshiba drive.  There are no SCSI Disk type error messages appearing.  It
is set up as the second of two scsi drives, with a scsi tapedrive on the
same bus.  disk0 is id 0, disk 1 is id 1, tape is id 4.  Drive 1 (the
toshiba) is external to the system.  I am sure the hardware config is
correct, and have no reason to doubt the disk itself, since no errors
are being generated.

The problem is that directories/files on that disk are getting badly
corrupted.  The whole disk is one partition, and it's being mounted on
/usr/spool/news (Esix gives me a warning when I do the mount).  It seemed
to go ok for a couple of days -- then "bad directory <>" messages
started appearing when I did 'du'.  Then processes accessing files/
directories on that partition started hanging, requiring a reboot to
kill them.  That's probably becuase the directory entries are trashed.

The filesystem type is S51K.  When I tried to use FFS, it caused files
both on BOTH disks to get corrupted!  With disk1 using S51K, disk0 
seems to be fine.

There is some question in my mind as to whether the disk is configured
correctly (the filesystems created correctly).  I used the Esix diskadd
utility (there is both a diskadd, and an adddisk -- diskadd calls addidsk)
to do the configuration.  Since I wanted more inodes, I then did a
mkfs blocks:64000 (I don't remember the number of blocks, but I used
whatever diskadd had given the disk).   /etc/partitions has at least
partly faulty data - it has the wrong number of heads listed.  What
is that used for?  Would that cause corruption?  Does it think it's
using a certain head to write, then not finding that head, using some
other head and clobbering data?  

I'd really appreciate hearing from anyone with a similar setup ...
BTW, I tried editing /etc/partitions, but then got errors about
"incorrect setup" or soemthign when I tried to run mkfs.  The documentation
is pretty sparse on this topic except for a few widely-dispersed man pages.

Thanks in advance!

If you can, please reply to bill at unixland.uucp;  if not, just a regular
reply to this address is OK.

Bill

-- 
Work: heiser at tdw201.ed.ray.com
      {decuac,necntc,uunet}!rayssd!tdw201.ed.ray.com!heiser
Home: bill at unixland.uucp, uunet!world!unixland!bill  Public Access Usenet/Unix
      (508)655-3848 12/24, 655-8723 96-HST, 651-8733 96-PEP/V32



More information about the Comp.unix.sysv386 mailing list