Hello there. There are many posts on this forum dealing with ODBC access problems and BE's catalog. I tried solutions from all of them (renaming catalogs folder and restarting, reindexing, running database repairs, etc.) with no success.
It turns out the database is not the problem - there is nothing wrong with the catalog structure or the database. The problem seems to be that there is some corrupted or duplicated data associated with particular media (disk or tape media, it makes no difference).
If this is your problem, when you attempt to inventory or catalog affected media, you will get the ODBC access error - and it won't matter what other repair steps you've taken, it'll always pop up with that media on that particular server (even if the media works fine in other servers). To correct this problem, you must make BE forget about the media completely (thereby deleting the bad information).
Here is the process.
1. Identify all media that causes the "ODBC access error" to appear when it is accessed.
2. Locate the entries for these media under "Media" in all media pools.
3. Move all of these affected media to the "Retired Media" pool.
4. Select and delete all of these affected media. (You can only delete "Retired Media")
5. Re-inventory and re-catalog the physical media. The ODBC problem should not reappear.
Once you have deleted all affected media out of the media pools, you should be able to run inventory and catalog jobs on those media again. If you find certain media is still problematic, you may need to repeat the above steps and then relabel the media. You'll lose the data, but get back the use of the media.
I have not been able to figure out why this happens yet - as far as I can tell, it's related to importing tapes, perhaps when multiple tapes are imported at once, or when certain kinds of unrecognized media (like old unix tar tapes) are put into the drives prior to relabeling. BE does a kind of fast preliminary scan of the media label as soon as it is inserted that is not part of inventory or catalog. It happened to me when relabeling media that was a mix of old BE 9.1 tapes and Unix tar tapes for use as new media under 10.1. My servers ran fine for months prior to this and seem to be running fine again after this fix.
Hopefully this will help some people with this problem.
