Re: [Exim] retry db owned by root

Philip Hazel wrote:

> That figures. It will most probably always be created by a root process
> for the first delivery that occurs.
>
> I suspect that the change I'm about to make will cover up the problem by
> getting the ownership right, but it may still leave you with an unusable
> db file if indeed it was a failure during the DBM opening process. We
> may have to insert extra debugging into the code to track it down.

Yesterday I talked to Kirill, and I had a look at the source.
Kirill told me that OBSD has the db-functions included in libc, so maybe
they are not completly the same as the BDB-functions.

This is just a guess, but is it possible, that O_RDONLY | O_CREAT does
not work? Like, if O_CREAT is set, O_WRONLY/O_RDWR must be set,
otherwise it fails?

ciao