RE: Ontape -r
This brings me to another question. Why does the restore process require
the lgoical logs to complete the restore process? Let me explain by giving
an example.
I did an ontape -r warm restore recently (after completing Y2K testing) and
when I responded No to the logical log restore question, it ended the
restore session stating that the restore was incomplete and left my chunks
in an offline status. What I wanted to do was just restore to the point of
the level-0 archive. Eventually I put in the tape for the logs and
completed the restore.
I suppose the detailed reasoning is in some chapter in the manual (so I'll
be re-reading it). Just looking for a short answer off the topp of your
experienced heads.
Best regards,
Edmund Nigel Gall
Senior Information Systems Specialist
Process Plant Services Limited
Atlantic Avenue, Point Lisas Industrial Estate
Point Lisas, Couva, Trinidad & Tobago, W.I.
Tel: (868) 636 2906 x 274 Fax: (868) 679 3770
-----Original Message-----
From: William Rice [mailto:ricew@operamail.com]
Sent: 1999 December 14 5:43 PM
To: Anne Shelp; informix-list@iiug.org
Subject: RE: Ontape -r
Did you roll forward all of the logical logs? If you do not do a complete
restore of the system I believe you are required to roll forward all of your
logical
logs to make them consistent with the rest of your instance.
Will
>===== Original Message From "Anne Shelp" <shean@mcdss.co.gov> =====
>Hi all,
>
>I was testing the capabilities of ontape to read an archive. I restored our
test database from the last tape. I took informix offline and ran ontape -r
develop. Message returned that the physical restore of develop completed.
The
online.log
>follows. The problem is now when I try to run ontape -s -L 0 I get the
error
"... Archive failed - DBspace develop is being recovered." What do I need to
do to fix this. I could not find the answer in the manuals.
>
>System info.
>
>IDS 7.20 Solaris 2.5.1Compaq proliant 4000
>
>Online.log:
>Sat Dec 11 12:17:36 1999
>
>12:17:36 INFORMIX-OnLine Stopped.>
>Sat Dec 11 12:20:02 1999
>
>12:20:02 Event alarms enabled. ALARMPROG =
'/usr/informix/etc/log_full.sh'
>12:20:04 DR: DRAUTO is 0 (Off)
>12:20:05 INFORMIX-OnLine Initialized -- Shared Memory Initialized.
>12:20:05 Physical Recovery Started.
>12:20:05 Physical Recovery Complete: 0 Pages Restored.
>12:20:05 Logical Recovery Started.
>12:20:08 Logical Recovery Complete.> 0 Committed, 0 Rolled Back, 0 Open, 0 Bad Locks
>
>12:20:08 Dataskip is now OFF for all dbspaces
>12:20:08 Quiescent Mode
>12:20:08 Checkpoint Completed: duration was 0 seconds.
>
>12:42:06 WARNING! Chunk 3 is being taken OFFLINE for recovery.
>12:42:06 Physical Restore of develop started.
>
>12:50:37 Checkpoint Completed: duration was 0 seconds.
>12:54:37 Physical Restore of develop Completed.
>12:54:38 Checkpoint Completed: duration was 0 seconds.
>12:59:51 INFORMIX-OnLine Stopped.>
>Sat Dec 11 13:45:03 1999
>
>13:45:03 Event alarms enabled. ALARMPROG =
'/usr/informix/etc/log_full.sh'
>13:45:06 DR: DRAUTO is 0 (Off)
>13:45:07 INFORMIX-OnLine Initialized -- Shared Memory Initialized.
>13:45:07 Physical Recovery Started.
>13:45:07 Physical Recovery Complete: 0 Pages Restored.
>13:45:07 Logical Recovery Started.
>13:45:10 Logical Recovery Complete.> 0 Committed, 0 Rolled Back, 0 Open, 0 Bad Locks
>
>13:45:10 Dataskip is now OFF for all dbspaces
>13:45:10 Quiescent Mode
>13:45:10 Checkpoint Completed: duration was 0 seconds.
>13:45:16 On-Line Mode
>13:45:16 Affinitied VP 3 to phys proc 3
>
>13:45:16 Affinitied VP 1 to phys proc 2>
>I wish I really knew what I was doing
>
>Anne Shelp
------------------------------------------------------------
This e-mail has been sent to you courtesy of OperaMail, as a free
service from
Opera Software, makers of the award-winning Web Browser, Opera. Visit us
at
http://www.opera.com/ or our portal at: http://www.myopera.com/ Your free
e-mail
account is waiting at: http://www.operamail.com/
------------------------------------------------------------
Nigel Gall wrote:
> This brings me to another question. Why does the restore process require
> the lgoical logs to complete the restore process? Let me explain by giving
> an example.
>
> I did an ontape -r warm restore recently (after completing Y2K testing) and
> when I responded No to the logical log restore question, it ended the
> restore session stating that the restore was incomplete and left my chunks
> in an offline status. What I wanted to do was just restore to the point of
> the level-0 archive. Eventually I put in the tape for the logs and
> completed the restore.
>
> I suppose the detailed reasoning is in some chapter in the manual (so I'll
> be re-reading it). Just looking for a short answer off the topp of your
> experienced heads.
[Previous post SNIPPED]
Remember the purpose of Informix's archive and restore procedures and utilities
Nigel Gall wrote:
> This brings me to another question. Why does the restore process require
> the lgoical logs to complete the restore process? Let me explain by giving
> an example.
>
> I did an ontape -r warm restore recently (after completing Y2K testing) and
> when I responded No to the logical log restore question, it ended the
> restore session stating that the restore was incomplete and left my chunks
> in an offline status. What I wanted to do was just restore to the point of
> the level-0 archive. Eventually I put in the tape for the logs and
> completed the restore.
>
> I suppose the detailed reasoning is in some chapter in the manual (so I'll
> be re-reading it). Just looking for a short answer off the topp of your
> experienced heads.
[Previous post SNIPPED]
Oops that last got away from me. Again:
Remember the purpose of Informix's archive and restore utilities and
procedures and it all makes sense. The purpose is to RESTORE the
engine, or significant pieces of same, to the state they were in at the
moment of the crash, or as closely as possible to do so. If this is the
goal, then restoring the archive without the logs is anathema.
HOWEVER, don't believe everything you read! What happens if the
engine crashes immediately after an archive is completed and there
has been no activity on the engine since the beginning of the archive?
In this case there would be no logical logs to restore! So? So, lie to
the engine! Say, as you did, NO there are no logs to restore. Then
ignore the dire warning about chunks being down etc. When the
restore ends the engine will be in quiescent mode if a cold restore was
being performed. Just bring the engine to full online mode WITHOUT
SHUTTING IT DOWN FIRST and the normal fast recovery
mechanism will kick in and mark all the chunks ONLINE. If you are
performing a warm restore just run onspaces -s .... and mark the chunks
online manually! After all, although the chunks that were restored are
not up-to-date they are internally consistent, they were restored from
a guaranteed consistent archive weren't they? So onspaces will
permit you to mark them up. You will only have a problem if you shut
the engine down after a cold restore before bringing the engine
ONLINE and waiting for the fast recovery to complete. Then the
chunks will be marked with status 'I' instead of the 'D' they will have
after a warm restore and onspaces will not help. Only restoring again
or calling tech support can change the status of those 'I'nconsistent
chunks.
Art S. Kagel