All...
On the 25/4/00 I posted the following regarding a failed V4.0f upgrade:
==================Original
posting============================================
Hi all...
Am doing an upgrade to V4.0f from V4.0D. The update started OK and got as
far as patch 1 - i.e.., all checks completed OK.
The system started reporting hard errors relating to the CD. I ended up
having to ctrl/c out of the update process.
I know the drive is OK but did suspect that the CD itself may have been
corrupt in some way.
I have another CD now. When I try to run installupdate, it fails with:
:root> /sbin/installupdate -i /dev/rz5c
Error:
I have rebooted the system (back to single user mode etc) but running
installupdate fails again - same really helpful error.
I have lost 17mb of /usr so something must have happened from the aborted
update. Does anybody have any ideas pleas? The system is currently down.
I don't want to have to do a full o/s restore (back to V4.0D) if I can avoid
it.
Regards - Tony
=================End of original
posting======================================
I had replies from:
Olle Eriksson [olle_at_cb.uu.se]
Mitch Kulberg [mok_at_unx.dec.com]
simon.millard_at_barclays.co.uk
Harald Baumgartner [hmb_at_mpe.mpg.de]
Paul Henderson [henderson_at_unx.dec.com]
Some asked for more information, but basically said that they had
experienced the same thing and that the only solution was to do a full
restore and to restart the upgrade from some known good media (obviously I
would not have started the upgrade if I had suspected the media was bad).
Luckily for me we have a comphrensive restoration procedure documented which
allowed me to go quickly into action and restore the o/s prior to repeating
the upgrade again.
Paul Henderson of DEC said he would pass this onto TRU64 engineering for
further comment. Having sent in the upgrade logs and console log extract, I
received a reply which basically told me nothing I didn't already know - but
many thanks for looking at the problem anyhow.
I do find the behaviour of setld rather bazaar - that it has not been
engineered to permit a recovery from an aborted upgrade. An update should
not fail in such a catastrophic way due to bad media.
The other DEC (sorry COMPAQ) O/S - the VMS one - handled upgrades properly.
i.e.., it built the upgrade in a set of temporary directories then, at the
last minute actually copied these over to the live system. Thus, if the
update failed for any reason (duff media, power failure etc) then you didn't
end up with a corrupt system.
I wonder if there are any plans to expand the functionality of installupdate
to that held by VMS 15 years ago?
Received on Tue May 02 2000 - 08:38:52 NZST