ADSM-L

[ADSM-L] Version 6 disaster recovery

2012-07-10 16:27:21
Subject: [ADSM-L] Version 6 disaster recovery
From: Thomas Denier <Thomas.Denier AT JEFFERSONHOSPITAL DOT ORG>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 10 Jul 2012 16:19:46 -0400
We have three 6.2.2.0 TSM server instances on two zSeries Linux systems
in two different locations. Each location has an instance for storing
client backup files, and one location has a dedicated library manager
instance. Each client file instance writes copy storage pools and
database snapshots to a tape library in a third location. The library
manager writes database snapshots to virtual volumes at the other
location. Each instance writes recovery plans to an instance at the other
location.

In the event of a disaster at one of the server locations, we plan to
recreate the lost instance or instances on the surviving Linux system.
I am attempting to create scripts and detailed instructions for this
process. Version 6 DRM is turning out to be a considerable disappointment.
As far as I can tell, IBM ripped out the parts used to create an empty
Version 5 instance and didn't bother to replace them with corresponding
Version 6 functionality, opting to dump the problem into the customers'
laps. I have so far not found any detailed discussion of empty instance
creation specific to 6.2, and and the only detailed discussion I have
found for 6.1 is logically inconsistent.

I am wondering if the process outlined below would work.

Preparations:
1.Define all three instance users on both Linux systems, with matching
  numerical values for user and group.
2.Include stanzas for all three TSM instances in
  /opt/tivoli/tsm/client/api/bin64/dsm.sys on both Linux systems.
3.Have each Linux system back up its instance directories to a TSM
  server at the other location.

Disaster recovery:
1.Restore the instance directory.
2.Replace the server options file, device configuration file, and
  volume history files with the corresponding files from the latest
  recovery plan.
3.Update the device configuration file, replacing all occurrences of
  the IP address of the destroyed Linux system with the IP address
  of the surviving system.
4.Run the database restore command from the recovery plan.
5.Update the IP address of the restored server in various TSM databases.

If the lost location were the one with the library manager, the library
manager would be the first instance recreated at the surviving location.

Thomas Denier,
Thomas Jefferson University Hospital
<Prev in Thread] Current Thread [Next in Thread>
  • [ADSM-L] Version 6 disaster recovery, Thomas Denier <=