Siebel repository migration

Question

Very new to siebel and I want to perform a repository migration from one environment to another. The command I am using is something like this on the target serve

./srvrupgwiz /m master_Test2Prod.ucf

so my question is what happened if repo migration fails in the middle and unable to continue?
Will the target environment becomes corrupted? Is there a way to recover?
I am thinking must be a way to take a backup of the current repository on the target environment and somehow be able to restore that?
If this is true, then how to do that?

thanks

Answer 1

By default, the Siebel Respository you are replacing in the target environment will be renamed to "SS Temp Siebel Respository". You are prompted to supply the name for the newly imported repository (which will default to "Siebel Repository"). When a new repository row is being imported, its ROW_ID value is appended to the end of the name you provided. Once it is successfully committed, that suffixed value is removed. Therefore you can always tell when a repository is partially imported. If something fails, it's perfectly safe to delete the partial one (or leave it there, the next attempt will result in an entirely new one with yet another ROW_ID value suffixed to the end). You can recover the old one simply by renaming it. You can see the exact steps followed by the Database Configuration utility's Migrate Repository process by looking in the UCF files that drive it (e.g. master_dev2prod.ucf and driver_dev2prod.ucf).

Answer 2

In all fairness Siebel version and Database system have little influence on the type of solution that most will put in place: which is reversal of the database changes.

Now, Oracle, Microsoft and IBM (only supported brands) each have their own approaches and I'm more familiar with those of Oracle. Many Oracle implementations support flashback. This is a rolling log of all changes which allows one to 'travel back in time' by undoing the statements. This includes deletes as well. The maximum size of this log is one to have attention for as the Siebel DB is quite a large volume of data to be imported. I'm sure that Microsoft and IBM systems have similar technologies.

In any case the old fashioned export to disk works in all systems.

Answer 3

You can backup the existing repository by going to Repository object type in the object explorer and renaming the existing repository in the siebel tools.

In case the repository import fails, you just need to change the name of the backed up repository to Siebel Repository.

Also use /l log_file_name in the command to capture the logs of the import process.

Answer 4

Your command is fine for a migration of repository using an answer file. However, you can split out the repository migration into individual commands rather than using the unattended upgrade wizard. One of these commands is (windows):

%SIEBSRVR_HOME%\bin\repimexp.exe

You can use this executable to import or export repositories. It is often used as a means to backup existing repositories, which tends to be referred to as "exprep". Rather than spend additional time during a release doing full export from source then import into target, the export from source can be done in advance writing out to a .dat file which represents the entire repository. This file can then be read in as part of a repository import which can save time.

In order to perform an export/backup of your current repository, you can use a command like below (windows):

%SIEBSRVR_HOME%\bin\repimexp.exe /A E /U SADMIN /P PASSWORD /C ENTERPRISE_DATASOURCENAME_DSN /D SIEBEL /R "Siebel Repository" /F c:\my_export.dat /V Y /L c:\my_exprep.log

Once you have the exported .dat file, you can run a repository import referring to this file, rather than a database with your repository inside. You do this the same way using an answer file like in your original command, but the answer file will reference the .dat file. You can step through the Siebel wizard in order to write out this answer file if you are not confident editing it manually.