>
> What version of Amanda?
>
Pretty much the latest, 242p2, I think.
>
> The netusage would not affect planner. And I don't think inparallel is
> involved, either. I'm pretty sure planner ships a sendsize request to
> all the clients at the same time and doesn't pay any attention to either
> of those limits.
>
> In theory, it has 1000 handles allocated (proto_init call in planner.c).
> Do you have some incredibly large disklist?
We have about 75 entries in the disklist.
It seems to have gotten into some kind of loop and chewed up all the
handles. Still not sure why. I changed a lot of parameters at the same
time, increasing 'inparallel', 'netusage', and 'runtapes' significantly.
Could any of those changes cause the problem?
I also moved to an 'incronly' strategy, because we're backing up over a WAN
and need to hold off on full backups until weekends. I'm not sure if it's
related, but through many trials, I figured out that if I add a new disk
with this incronly strategy, I need to force a full backup right away, or
else I get "no results" back from the "getting estimates" phase, and the
backup fails.
It's running again now, but I had to blow away the curinfo database and
rerun a full backup of everything to get it to work.
Regards,
Carey
>
> >Carey
>
> John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
>
