SCSI disk sources

Farhad Shakeri farhad at CS.Stanford.EDU
Sat Apr 27 10:00:52 AEST 1991


In article <1991Apr22.191208.796 at roundup.crhc.uiuc.edu>, steven at pacific.csl.uiuc.edu (Steven Parkes) writes:
|> We're in the market for a 1+GB SCSI disk and an Exabyte 8200.  So far we have
|> quotes on the Fujitsu M2266, the WREN VII, and an HP (I don't have the model
|> number at hand.)  Any notable observations on these or other drives would be
|> appreciated -- things such as noise, reliability, and performance.
|> 
|> We are also looking for sources that have good (university) pricing and have
|> currently talked with NPI(IL), R^2(CO), and Datalink(IL).

Well I was about to post a question  and I saw this related
article: (I hope it will help Steven too)

I have a Fujitsu M2266s drive and I have a major problem if  I was to 
use the entire disk as one partition (c=2107704 / 512  given by "chpt -q
device-name " ).  The problem is that, if I fill the disk %111 then I lose
the entire filesystem.  I get the error . or .. not found and fsck gives
me the famous :

	BAD SUPER BLOCK: MAGIC NUMBER WRONG
	USE -b OPTION TO FSCK TO SPECIFY LOCATION OF AN ALTERNATE
	SUPER-BLOCK TO SUPPLY NEEDED INFORMATION; SEE fsck(8).

I could not recover anything with any alternate super-block.


I was told by another programmer that the problem is with the Ultrix 4.1
scsi driver that can not handle more that one GIG. (old ansi standard?).
Well to fix this problem I decreased the size of c partition to
(1705*15*82=2097150)  this has fixed the problem with not much loss.
  ^   ^   ^    ^
  c   h  s/t   <------These numbers were given in R-Squered Manual

Ok my question: Is there a fix for the scsi driver under 4.1A or
is 4.2 going to have this fix?  Has anyone seen this problem?

I think Ultrix 4.1A scsi driver is still in group 0 of American National
standard for scsi and what I need is a scsi driver in  Group 1.
for more info you can look at ANSI X3.131-1986 page 95 for group 0
Table 8-10  and page 116 for group 1 Table 8-26 .

Oh the price we paid for M2266S-512 from R-Squered was below $3000.

Noise level was high when I was doing heavy IO (5 dds of bs=100k).
In normal performance in a quiet room + a DS3100, it sounds like a
chicken once in a while.  :-)  but much better than rz55.
I have not done a real performance test, maybe someone else can help
you on that.

Thanks in advance!

-- 
       +----------------------------------------------------+
      /   Farhad  Shakeri         E-Mail:                  /
     /  Stanford   University     farhad at CS.Stanford.EDU  /
    / Computer Science Department                        /
   +----------------------------------------------------+



More information about the Comp.unix.ultrix mailing list