Do You Know SCO? (clarification)
richard childers / kg6hac
fscked at pacbell.net
Thu Oct 9 15:58:33 PDT 2003
I'm seeing a lot of good answers, but it's also clear pretty much
everyone thus far has about as much SCO experience as I do, and so the
suggestions are fairly common-sense but lacking in those gritty details.
(format.dat's a good guess, though; tip of the hat to Maureen Woodward,
she's in the lead, such as it is.)
It might help if I clarify matters by saying that it is -not- the boot
disk which was replaced.
Let's see, I think I have some error messages scribbled down here ...
haven't Google'd 'em yet.
G hd_config
WARNING: hd: no root disk controller was found
H init ime Loadable Driver may be required
G drain8042
PANIC: srmountfun Error 19 mounting rootdev (1/42)
Error 19 opening dumpdev (1/41)
Dump not completed
*** Safe to Power Off ***
I also note that the devices were all visible to the SCSI controller's
BIOS, so that shoots my hardware incompatibility theory down ...
(Gee ... I could see how printing a cryptic error message on the
whiteboard, and then challenging the crowd to diagnose the problem, one
question at a time, with a prize for whoever gets it first, could be a
real crowd-pleaser ... at a meeting of systems administrators.)
-- richard
richard childers / kg6hac wrote:
> Do you know SCO?
>
> Santa Cruz Operations' UNIX, that is?
>
> I don't know the version of SCO; but I know someone, who has a
> customer, whose SCO installation had a hard drive die. They need help.
>
> It would be fair to charge $80 an hour - but it would also be fair to
> say that you would be expected to provide expertise, in exchange for
> the privilege of charging money, for sitting on your butt, restoring
> their data.
>
> Before I can provide interested parties with the contact information,
> you must convince -me- that you know more than I do about this problem
> ... so that I can hand it off to you, in good faith.
>
> Here's the history.
>
> I was brought in to diagnose the problem; hardware or software? After
> some poking around, I found SCO's equivalent of /usr/adm/messages, and
> in it, hard error messages naming a device and the bad blocks. I
> extracted the device name, translated the UNIX nomenclature into a
> SCSI device for the benefit of the service provider, and also provided
> a list of bad blocks.
>
> Several weeks later, I was called and asked to come down ASAP; the
> service provider was returning the computer to the customer, the drive
> had been swapped, but it wasn't booting with the new drive, only when
> they put the old drive in. He was hoping I could help.
>
> The old drive was a once-widely recognized 9 GB SCSI of, I'd guess,
> late 1990s vintage. He was unable to replace it, he said, and so had
> installed a larger 30 GB drive, but it didn't work.
>
> Although I won't go into the physical hardware here, it's not unlikely
> that there was a mismatch in the SCSI controller, cable, and drive; I
> have recommended that he get an identical drive simply because
> although it is very intellectually satisfying to puzzle out what is
> not working, it also takes many hours, is rarely cost-effective, and
> requires technicians with at least a two-year degree from a junior
> college in electronics or computer hardware, which, I judge, may be
> lacking here (and for this reason, diplomacy is required, as well as
> technical expertise).
>
> To make matters worse, throughout this sequence of events he had
> insisted that Norton Ghost would be sufficient unto all his
> requirements. Having recently dealt with a Linux server that had been
> Ghosted (and it was one un-bootable, un-filesystem-checkable piece of
> silicon, too), I had my doubts; I know too much about how filesystems
> and hard drives and bad block mapping works to assume that you can buy
> a one-size-fits-all solution, shrinkwrapped, at CompUSA.
>
> But I have not said so explicitly; I hate to argue with a customer.
> (Again, diplomacy is indicated.)
>
> My intuition is that there are a combination of problems here; a
> hardware mismatch combined with an operating system gotcha.
>
> (I'm not against working on this directly, if there's someone out
> there who wants to answer questions and permit me to do the dirty work
> in exchange for a small honorarium for their wisdom and guidance, by
> the way, if they're reading this, are knowledgeable, but are feeling
> undermotivated ... contact me.)
>
> My intuition is that the solution involves negotiating the hardware
> mismatch ... reconciling the operating system gotcha ... building a
> new filesystem ... and reloading the data from backups, made either
> immediately before the disk is exchanged, or from a stack of tapes
> whose contents and dates of creation would probably need to be audited
> and verified beforehand (IE, more hours of work).
>
> If you are reading this, have detailed expertise with SCO, and can
> provide objective references to materials which corroborate your
> diagnosis, analysis, or terminology (IE, URLs), I'd like to discuss
> either leveraging off of your expertise, or handing this off to you,
> entirely, so that the customer - poor wretches, they are underpaid,
> overeducated financial and statistical analysts trapped in a world of
> high finance but receiving only crumbs for their economic labors ...
> they have more in common with systems administrators then they do with
> financiers - can have their computer back, working reliably, their
> faith in UNIX (dare I say it?) restore(8)'d.
>
> (-:
>
> Thanks in advance,
>
>
> -- richard
>
> Richard Childers / Senor Engineer
> Daemonized Networking Systems
> https://www.daemonized.com
> (415) 759-5571
>
>
>
> ----------------------------------------------------------------------------
>
> You have received this message via the baylisa-jobs at baylisa.org
> mailing list.
> For questions or comments regarding list admin, email
> postmaster at baylisa.org.
>
More information about the Baylisa
mailing list