Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance
From: IanC (ian.nospam_at_intech.co.uk)
Date: 05/21/04
- Next message: Bela Lubkin: "Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Previous message: Brian: "Re: IBM speaks by memorandum..."
- In reply to: Bela Lubkin: "updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Next in thread: Bela Lubkin: "Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Reply: Bela Lubkin: "Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Fri, 21 May 2004 08:34:23 +0100
"Bela Lubkin" <belal@sco.com> wrote in message
news:20040520185944.GW10272@sco.com...
cut
> > .
> > Thanks Bela
> >
> > I have now loaded the new driver.
> >
> > I links OK
> >
> > It boots OK
> >
> > The backup job used to take 22 minutes.
> >
> > It now takes 99 seconds - wow
>
> Excellent!
>
> > This is on a fairly modern PC with a UDMA 7 drive. I expect that the
> > performance improvement will not be as great on earlier machines.
>
> You're probably right. OSR5 is fairly conservative about which chipsets
> it will attempt UDMA on. The older machines will also contain slower
> drives and slower controllers (in some random combination that will
> often mismatch fast drive / slow controller or vice versa...) But it
> will probably make a striking difference on a good proportion of them.
>
> > How 'safe' do you think it will be to use this on older machines ?
>
> You should run your first test machine this way for a few weeks first,
> in case there's something you haven't noticed. You should also be sure
> to have good backups of each machine you modify -- this is uncharted
> territority and you shouldn't act like it's totally safe, even if it
> turns out to be.
>
> The driver is _intended_ to work correctly on all hardware it's ever
> worked on. If something goes wrong it won't be because code to support
> old quirky hardware has been intentionally removed or repurposed. But
> mistakes are always possible.
>
> > I will need to automate the installation in order to get it to 250 sites
> > without driving around with a floppy disk - any ideas?
>
> Yes: `btldinstall` doesn't actually care whether the stuff it's
> installing is on an actual filesystem. If you mount the "wd BTLD"
> image, you can save off the key parts:
>
> mount ... /mnt
> cd /mnt
> tar cf - wd wd506 install | compress -H > /tmp/wd.btld.Z
cut
>
> You must include the contents of ./wd as well as ./wd506, even if you
> don't have any OSR507 systems in the mix. For many files, ./wd contains
> the real meat and ./wd506 just a reference to the real file in ./wd.
Thanks again Bela
Before I received your reply I had already got as far as tar'ing the /mnt
directories from the floppy.
I did an instalation on a second test system but I had only copied the wd506
directory and not /mnt/install or /mnt/wd.
It appeared to run btldinstall correctly = no error messages and the machine
has speeded up.
Will that be OK?
We will proceed, as always, with caution onto a number of test systems,
pilot sites etc
Cheers
IanC
>
>
============================================================================
=
>
> I haven't addressed OSR500. If someone wants to try this procedure on a
> 5.0.0 system and report back to me, I can probably repair whatever goes
> wrong. I suspect the kernel won't link; show me the error messages...
>
> >Bela<
- Next message: Bela Lubkin: "Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Previous message: Brian: "Re: IBM speaks by memorandum..."
- In reply to: Bela Lubkin: "updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Next in thread: Bela Lubkin: "Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Reply: Bela Lubkin: "Re: updating "wd" driver on OSR502 and newer, Re: Hard Disk Performance"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]