Retrying is not always possible, given what might occur during a process. The file that the process pertains to may no longer be available after a process has executed, or perhaps has been replaced by a new file.
In addition, there may be script in the process or constituent projects that may make a retry operation impossible.
You can set up alerts so that you receive notification via e-mail, or write specific information to the event log that can then be picked up by any network management systems you have in place such as MOM, Openview, Tivoli etc.
In most cases, these types of failures should involve user intervention in their resolution.
If you have certain well-known issues such as an unreliable database or network, then these could be checked at process inception time by writing pre-process script, so that the process could retry until the issue has been resolved downstream, then start. This will not cover all eventualities, but may be a useful measure.
Is it specifically an unreliable database that is an issue, or are there other issues that you have as well?
We can certainly try and see if there are scripts we could write that might be helpful, in addition to considering certain failure scenarios that may be recoverable for a future release.
Let me know a bit more about your issues.
However, I don't see any way to get it to auto-retry a job that fails part-way through an export to SQL Server if, say, SQL Server goes down. I get an error in the log file and that's about it.
This really seems like a significant limitation in MDP in that it's not very robust in the face of job failures due to database server issues, etc.
When I look at the failed job inside the MDP snap-in, it also doesn't give me any option to retry the job, which also seems suboptimal.
It looks like my only recovery options are to either manually restart the process, or build some kind of custom application that would monitor MDP for failed jobs and restart them....but then why did I just pay $10,000+ for this product that's supposed to handle that sort of thing???
SQL Server Agent has options to retry jobs when they fail--why doesn't MDP?
If I'm missing something obvious here, please point me in the right direction.[/QUOTE]