T.R | Title | User | Personal Name | Date | Lines |
---|
218.1 | Hello? | KERNEL::SMITHERSJ | Living on the culinary edge.... | Mon Mar 16 1992 09:18 | 3 |
| Help anyone?
Thanks
|
218.2 | I suppose that puts me on the spot | IOSG::TALLETT | Mit Schuh bish hi | Tue Mar 17 1992 09:19 | 15 |
| Hi Julia!
Oh dear, my worst nightmare, someone asking how to repair a
failed transfer user! I don't know what I'd do if I were in
the field and a customer bought a days consulting to fix such
a problem! Go off sick maybe! :-)
Actually I was hoping one of those kind people in Atlanta would
post the "Here's how to recover from a failed TU" article from
STARS. I don't know a lot about recovering from this situation
except BACKUP (but I never understood how you recovered the
shared areas...).
Regards,
Paul
|
218.3 | tricky one | IOSG::TYLDESLEY | | Tue Mar 17 1992 09:53 | 19 |
| ..just to endorse Paul's comments, Julia, a failed V2.4 transfer
is very difficult to deal with. We had one here (the account of a
highly-respected ex-ALL-IN-1 developer!) that went wrong (house-
keeping shut down ALL-IN-1 half way through the transfer). After
many attempts we resorted in the end to BACKUP and start again.
The account was 2-3 weeks out-of date from the backup, but as he
was already moved onto the target system (and using VMSmail), this
didn't seem to matter.
How difficult would it be for your customer to restore the account
and start again?
One other source of possible help - I seem to recall an article in the
old ALL-IN-1 conference, I think by Terry Porter, that gave more
details on how to salvage a V2.4 semi-transferred account.
In V3.0, thanks to Paul T's changes, Transfer is recoverable.
DaveT
|
218.4 | Haven't given up yet | KERNEL::SMITHERSJ | Living on the culinary edge.... | Tue Mar 17 1992 16:40 | 12 |
| Thanks for the moral support, lads! At least I'm not on site
so they can't beat me up. I have been trying out the amended
script in 'How to restore user to original status after 2.3 PAT'
which works OK for me to de-archive the files. I've given it
to the customer to try out. If (and oh boy, IF, it works),
I'll advise the customer just to transfer the ALL-IN-1
subdirectory rather than the VMS directory.
Keep your fingers crossed.
julia
|
218.5 | UAF wrong? | IOSG::TALLETT | Mit Schuh bish hi | Wed Mar 18 1992 08:06 | 7 |
|
Oh yes, we never got to your original problem. Sounds to me
as if the UAF doesn't have the device/directory in it?
(No, it doesn't sound convincing to me either,....)
Regards,
Paul
|
218.6 | Stars recovery does work | TRCOA::HALSEY | I'd rather be sailing! | Wed Apr 29 1992 17:52 | 20 |
| Sorry for a late reply, but I'm catching up on my notes.
At my regular customer site, we've got a fair bit of experience with
User Transfers and recovering from failed transfers. We have even
"documented" the procedure, its cautions AND how to recover from it!
(I can fax you a copy if you like).
The recovery is based on the STARS article provided by the local CSC.
It actually does work, although we made a subtle minor improvement.
It's based on two files which you can create RESTORE.SCP and RESTORE_
TRANSFER_USER_DOCUMENTS.SCP.
The only thing we haven't worked on is CTN. We just use a rule NEVER
EVER EVER use CTN. Blow the process away if you must, but don't use
CTN. This irrational fear is due to inexperience and rumours about
CTN, not from any actual use of it.
Hope it helps...
Bob Halsey, Toronto SWS
|
218.7 | It's worse than that Jim... | UPROAR::WATSONR | Dunno man... just got here myself ! | Fri May 08 1992 18:08 | 39 |
|
Oh no... I've just tried to PAT a manager's account and have totally
destroyed it !!! Moral support would be nice, advice would be better.
What I foolishly did was this... I started four PAT procedures for four
users. Silly I know, but I'd never done this before. Foolish bravado if
you will, but it gets worse... much worse. One works... two fail when the
disk fills up so... in a rash attempt to create enough space to save the
fourth, I delete the OA$TRANSFER area for one of the failed accounts.
The fourth then fails. I then create another OA$TRANSFER on a totally
empty disk, and restart the PAT process for the three failed accounts.
I then dutifully clean up my 'old' OA$TRANSFER area and await the
completion of the new PATs. One fails with an accvio. In the process of
trying to find out why, the whole horrible truth becomes apparent and
I realise what I've done. I immediately kill the other two PATs and
contemplate killing myself.
I think I'll kill myself now and save someone the trouble.
I've spent the last hour in the depths of dispair but have decided to view
this as a challenge. My plan is this... comments please.
1. Rename the user's accounts somewhere safe
2. Rename the OA$TRANSFER area somewhere safe
3. Restore the shared directories from last nights backup thus creating
any files that have been deleted.
4. Restore the user's directories.
5. Restart the PATs (one by one)
Is this a plan ? will it work ? Shall I hide somewhere dark and hope it all
goes away ?
:-(
Ross
|
218.8 | | UPROAR::WATSONR | Dunno man... just got here myself ! | Fri May 08 1992 22:48 | 24 |
|
� 3. Restore the shared directories from last nights backup thus creating
� any files that have been deleted.
�
� 4. Restore the user's directories.
Having checked a shared file that was restored, it would appear that
the shared DAF files have been updated by the PAT, removing entries with a
usage count of zero if the file that goes with the entry has been archived
by the PAT. Damn... that adds another level of complexity. Looks like I'll
have to shut down ALL-IN-1, move the shared area DAFs somewhere safe,
restore the old DAFs from last nights backup, THEN run the PATs and THEN
put the 'real' shared DAFs back again and restart ALL-IN-1.
This 'should' work because to all intents and purposes I'll end up with an
ALL-IN-1 system as it was last night. Nothing happened to the three accounts
in question between the backup and the PAT so that's OK as well.
I'm beginning to feel a BIT more confidant, of course that could all go out
the window if this doesn't work - which it will. Think positive Ross, this
IS going to work. Isn't it ?
Damn... this account transfer stuff is a bitch when it goes wrong (or rather
the user buggers it up).
|
218.9 | Usual plug - V3.0 PAT is better | IOSG::TALLETT | Making broccoli | Wed May 20 1992 20:09 | 13 |
| Hi there!
Sorry for the slow response, I was out at DECUS. Boy you sound
to be having fun!
All I can say is, V3.0 won't screw your source account. We got
beaten up so badly by people screwing up accounts with PAT that
we changed PAT to be readonly on the source account...
Your disaster sounds quite the worst I've heard of...
Regards,
Paul
|