I want to thank those of you who did respond:
Patrick O'Brien pobrien_at_cfa.harvard.edu
Stuart Davidson stuart.davidson_at_eurocontrol.be
Although, as Mr. O'Brien stated .5G ought to be large enough for this reconfig
-- in this particular environment it was 40,000 block too little to even load
the OSF realtime subsets on top of the current configuration. (This system was
configured to download 3 heterogeneous systems running VXWorks and having no
disk space themselves.)
For any of you 'brave' (very brave) souls who ever want (need--no one wants to
configure a system this way) the following is a summary of what I actually went
through. (It was not that horrific, but it was a different experience.)
Planned to proceed as follows:
CORRECTIONS/DISCREPANCIES ARE NOTED IN CAPTIAL LETTERS UNDERNEATH EACH STEP
1) dump|restore the current kernel (actually both / and /usr) to the
hard drive
2) install more of the optional subsets (realtime kernel and programming
subsets)
OSFRTBIN200, OSFRTBIN200, and OSFRTDEV200 were installed
3) generate the new kernel
COMPILED A TEST PROGRAM USING THE REALTIME SUBSETS ON THE NEW
KERNEL
COPIED THIS EXECUTABLE TO A SET OF OPTICALS USING THE OLD KERNEL
AND ATTEMPTED (UNSUCCESSFULLY -- AS EXPECTED) TO RUN THE PROGRAM
4) de-install the following OSF subsets
OSFBIN200, standard kernel objects
OSFBINCOM200, kernel header and common files
OSFHWBIN200, hardware kernel objects
OSFRTBIN200 &
OSFRTHWBIN200, newly installed realtime kernel objects
ATTEMPTED (UNSUCCESSFULLY -- UNEXPECTEDLY) TO RE-COMPILE THE
TEST PROGRAM
REINSTALLED OSFBIN200 AND OSFBINCOM200 AND ATTEMPTED
(SUCCESSFULLY) TO RE-COMPILE THE TEST PROGRAM
5) dump|restore the new kernel to the optical disk
COPIED THE EXECUTEABLES COMPILED IN STEPS 3 AND 4 TO THIS
DISK AND RAN BOTH SUCCESSFULLY
All in all, it was a great success. An attempt at generating this new kernel
prior to starting on this path was to move /sys files to a second disk and put
in a soft link in order to get room on the / disk. The kernel would not
generate this way. I have fellow managers (who manage CRAY systems) who insist
this should have worked, although they have not done so themselves. Is this
just a theory or tribal legend gone bad....I will probably never know.
By the way -- only a slight graze to the scalp on this one and the foot is
healing fine.
Good luck to all.
Bonnie DeBruler
debruler_at_eglin.af.mil
======================================================
==//===
// // TYBRIN Corporation Corp. (904) 651-1150
\\===================================================
1283 N. Eglin Pkwy. Office (904) 882-9794
Shalimar, FL 32579 Fax (904) 729-2550
On Thu, 6 Jun 1996, DEBRULER, BONNIE wrote:
> Hello, fellow managers....
>
> I am getting very very brave these days.
>
> I am in the process of building a new OSF kernel on a large hard drive
> which will then be dump|restored to a very small (.5G) optical disk. Here
> is what I am planning to do, please let me know if there are things that
> will bite me later.
>
> 1) dump|restore the current kernel (actually both / and /usr) to
the hard
> drive
>
> 2) install more of the optional subsets (realtime kernel and
programming
> subsets)
>
> 3) generate the new kernel
>
> 4) de-install the following OSF subsets
>
> OSFBIN200, standard kernel objects
> OSFBINCOM200, kernel header and common files
> OSFHWBIN200, hardware kernel objects
> OSFRTBIN200 &
> OSFRTHWBIN200, newly installed realtime kernel objects
>
> 5) dump|restore the new kernle to the optical drive
>
> I realize that any future kernel builds would require these subsets, but I am
> planning on have the hard drive as a permenant 'master' of the o/s.
>
> Have I missed anything and will this shoot me in the head later? (foot's been
> blasted off long ago!) Has anyone had any experiences with this type of
> environment/administration nightmare that I should be aware of?
>
> As usual, I thank you in advance for any inputs.
>
> Have a good day )
>
> Bonnie DeBruler
> debruler_at_eglin.af.mil
> ======================================================
> ==//===
> // // TYBRIN Corporation Corp. (904) 651-1150
> \\===================================================
> 1283 N. Eglin Pkwy. Office (904) 882-9794
> Shalimar, FL 32579 Fax (904) 729-2550
>
>
Received on Mon Jun 17 1996 - 18:00:12 NZST