> You're not actually causing a sweep with the backup, just a garbage
> collect pass. Any non-indexed access to the tables has the same

adding the "select count(*) from table" before a activate the indexes
didn't help. I still have an uncontrolled timeout in the backup
process when it's running on this table. The first 1'000'000 records
are backuped in reasonable time, then the backup task's seems to stop
for ever and something is going on (heavy disc access).

So it looks like the select statement is not enough.

cu Christian