we've run into some problems with rpm when using Berkeley DB >= 5.3 as its backend:
An analysis of what rpm is trying to do, and where things go wrong:
DB 5.2.42 does the right thing (or at least what we expect to be the right thing in rpm).
We have a workaround for rpm that essentially confirms that the blob size returned by DB_NEXTDUP is at fault:
Any chance for a fix for the problem? Or are we doing anything wrong?
I have looked into this, and I have been unable to reproduce the error and cannot find anything in the code that could explain what you are seeing. Is it possible you could post the code in the functions db3cget() and db3cpget(), and post the configuration of the primary and secondary databases? If you would rather email that information to me I can be reached at lauren dot foutz at this company.