System Restore - Sun Solaris Client - Using Release 5.3.2

[Perry_Custer]

I wanted to check with others doing full system restores on Sun Solaris servers. What do you have to absolutely stay away from if you are restoring the entire server? If you have most of the system, but are not sure what is good and what is not, and you did not have a ufsrestore for instance. Would you stay away from '/' itself, but restore most of the directories below it except '/usr' which is also identified in the system restore documents. Is there any reason you can not restore everything else if you have a running system and then reboot at the end of the restore to get everything sync'd. I am aware this is not the proper or documented method where you use ufsrestore and start at that point. Even with that method it indicates not to restore '/' or '/usr' what about the other directories directly under '/'. I will take any pointers that can be provided, or previous testing with Full system restores on Sun Solaris 9 servers.

 

[chad_small]

You can restore all filesystems but the key is the options you use. You can restore / and /usr but you'll want to use the -REPLACE=NO option on those filesystems. Scripting the restore works better for this than doing it with the GUI since each filesystem will use its own command within the script. By doing this you can restore files that possibly were not part of the initial reinstall of the OS. I assume you are using jumpstart to rebuild the server? If so then its like a base OS install and you can definitely benefit from restoring those filesystems with the stated options.

 

[Perry_Custer]

Chad would you have an example of a script to use? Assuming you are talking about a kornshell script or a perl script? Could you provide a sample script? I would greatly appreciate it. Also what about the directories under '/' except for '/usr' are they all ok to restore without the -replace option - example: /opt or /sbin ?

 

[chad_small]

Perry,



I would restore all system filesystems with the -REPLACE=NO option. This will keep you from any problems and get you as close to back to normal as possible. The script I use is actually a simple Korn shell script that outputs to a file so I can monitor the restore. It's based of of the example Tivoli gives in its documentation. The file can get big so be careful.



#!/bin/ksh



exec > /tmp/tsmrestore.out 2>&1



dsmc res / -subdir -yes -replace=no

dsmc res /usr/ -subdir=yes -replace=no

dsmc res /sbin/ -subdir=yes -replace=no

dsmc res /opt/ -subdir=yes -replace=no

dsmc res /oracle/ -subdir=yes -replace=all





Once you get past the system dirs then you can start replacing files. I am sure there are better scripts out there this one just works for me and is easy to manage. Also, if you think you have the data on various tapes you could try and run multiple restores at the same time just be aware if there is tape contention one will wait til it gets the tape from the other restore.