Hi heLen,

> The only other idea that comes to mind, is that you have a

filesystem

> backup locking a sector of the database file at the time when sweep

wants

> to run.
>
> Other than that, I'm out of ideas.

Can it be that the sweep is being blocked by an ongoing transaction?
This problem seems to occur only when one or more connection(s)
is/are open to the database...

I am really not that familiar with how the sweep-mechanism work, but
I read somewhere that the "lucky" transaction that got to be the one
that triggered the sweep would hang until the sweep was done. This
could explain the hangs I experience, but on the other hand - in the
multithreaded version of the server software, the sweep runs in a
separate thread (hence the error message), so I guess this is not the
case after all... !?

When the sweep keeps failing, connecting to the database (or any
other database on that particular server) takes forever, and
eventually times out claiming that it could not access the password
database...

Additionally all the other tools (gfix, gstat, etc) hangs.