home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!mcsun!sun4nl!alchemy!edwin
- From: edwin@cs.ruu.nl (Edwin Kremer)
- Subject: Re: Trouble with HP disk on 4D/35 (long)
- Sender: network-news@cs.ruu.nl
- Message-ID: <1992Jul23.124707.25796@cs.ruu.nl>
- Date: Thu, 23 Jul 1992 12:47:07 GMT
- References: <l56ltrINNfcb@news.bbn.com>
- Organization: Utrecht University, Dept. of Computer Science
- Lines: 48
-
- In <l56ltrINNfcb@news.bbn.com> owczarek@izar.bbn.com (David Owczarek) writes:
-
- | I am having problems with an HP-97560-300 0B16 SCSI disk drive.
-
- Hmm, where have I seen that disk before...
-
- We bought almost the same disk (revision level was 0B11) via a 3rd party
- vendor and... hooked it up to a HP9000/433s running HP-UX 8.0. Here's the
- story...
-
- After hooking up the disk, a 'newfs' (it's like 'mkfs') using a (HP
- unsupported) disktab entry succeeded without a problem; so did an 'fsck'
- thereafter. However, when trying to mount the filesystem, mount complained
- with:
- /dev/dsk/3s0: unrecognized file system
-
- I repeated this sequence of commands with the same disk on a 9000/750
- running HP-UX 8.05 and after doing a 'newfs' there, the former machine
- was able to mount the file system (!?!). By then, I wasn't too confident
- about it, and I was right. This is where it all went real bad... after a
- couple of days the disk on the 9000/433s filled up to appr. 500Mbyte
- (still leaving more than 500Mb free) and strange things started to happen:
-
- - various programs complained about the filesystem being full, some others
- didn't complain but created zero-length files.
- - a 'find some-dir -atime +10 -exec rm -f {} \;' done by one of the users
- in order to cleanup the subdirectory 'some-dir' probably got the i-node
- wrong and... removed his $HOME -- yes I checked this carefully, it was
- NOT the user's mistake!!
-
-
-
- After all, we returned the disk. It's really very minor and obscure details
- with disk drives that seem to cause havoc. HP currently supports almost the
- same disk: HP-97650, rev. level 0BW2. I'm told it's really very low-level
- effects that break certain disk, like SCSI cable length; and you probably
- need lots of hardware experience together with a good oscilloscope a TDR
- and a fine logic analyzer to figure out what the real problem is.
-
-
- So far for my horror story, albeit on a different architecture...
-
-
- --[ Edwin ]--
- --
- Edwin Kremer, systems and network administrator. [NIC-Whois handle: EHK3]
- Department of Computer Science, Utrecht University, The Netherlands
- Email: edwin@cs.ruu.nl NeXT-mail: edwin@edge.cs.ruu.nl
-