4.0f cdfs bug

From: McCracken, Denise (Denise.McCracken@misyshealthcare.com)
Date: Thu Oct 24 2002 - 13:07:59 EDT


        I'm looking for a way to get around this problem. We have found
that until the cdfs.mod patch is installed, filenames that are copied from
CDs come up with file dates of December 1969. It's easy enough to patch
this after the OS is installed, but by then we have these filedates all over
the system from the install.

        What can I do to fix this? Someone suggested burning an install CD
with a patched kernel on it, but I'm afraid that this might affect the
patches, or the script that we use to install subsets after the base OS is
there. Any other ideas?

thanks

Denise McCracken
MiSys Healthcare Systems
Denise.McCracken@misyshealthcare.com
(520) 570-2521

-----Original Message-----
From: Jonathan Williams [mailto:jonathw@shubertorg.com]
Sent: Tuesday, October 15, 2002 2:24 PM
To: tru64-unix-managers@ornl.gov
Subject: SUMMARY: HSZ70 vs HSG80

Ok...restore problem solved!! Thanks go to Cory Erickson for suggesting I
look
into the writeback_cache on the HSZ70 controller. Sure enough, it was set
to
"nowriteback_cache". Once I changed it, the disks INSTANTLY jumped up to
5000KB/sec (which may be slow still, but at least it's on par with the
production machines).

Big thanks go to Charles Ballowe as well for suggesting I use "collect" to
monitor the restores in the first place. What a wonderful program!!

Thanks to everyone else who responded as well.

Jonathan Williams
Unix Systems Administrator
The Shubert Organization, Inc.

---Original Message---
> Tru64 5.1, patch 3, Alpha ES40 machines.
>
> I have a test machine here, and the only difference between it and our
> production machines is that the production machines are on a SAN with
HSG80
> controllers, and the test machine is not on a SAN and has it's own HSZ70
disk
> cabinet.
>
> I'm trying to track down an informix restore slowness we've been noticing
on
> this test machine. I've been using 'collect' to capture data during
various
> restore and backup operations. I've come to the conclusion that it is
> definitely a disk issue, and only with writing to the disks. When
comparing
> this test machine to a real production machine, all the numbers are pretty
> close, except for a restore operation that writes to disks out on an HSZ70
> controller. On the HSZ70, it seems to max out writing to disk at about
2500
> KB/sec, while on our other systems that use an HSG80 write at about 5000
KB/sec
> (which would explain why the restores take twice as long on this system).
> Reading from the HSZ70 disks is much like the production machines at
around
5000
> KB/sec.
>
> So I'm wondering, is this normal behaviour for an HSZ70? To write to it's
disks
> at around 2500 KB/sec...or is there a problem? I'm hoping that it's just
a
> difference between the two types of controllers. I've been looking at
some of
> the manuals, but can't find anything that seems too helpful.
>
> Oh, and both the HSZ70 and the HSG80 use the same type of disks
(DS-RZ1FC-VW
> 36.4GB).
>
> TIA



This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:48:57 EDT