> Is it normal that the "sendbackup" starts more than 1 hour later then
> the sendsize?   The estimate time is only a few minutes according to
> your report below.

No, this is not normal. I noticed that my server wasn't just 3 minutes
ahead of my client...it was 1 hour and 3 minutes ahead of my client.
This could be the root of all evil. I spent the morning syncing all of
my times, and hopefully this will eliminate my time problem. Maybe the
two just drifted too far apart to communicate effectively?
> 
> And then something completely different.
> Just a problem to investigate... (because it really strange!) 
> How big is your holding disk? Is it big enough to hold all 
> your images at once? The failing system generates a backup of 
> almost 5.5 GB. I know amanda tries to avoid the following 
> situation, but it could run into (and I'm not sure how amanda 
> 2.4.2p2 reacts) the situation that it is dumping a DLE to 
> disk, and the holdingdisk gets full. In that case the dumps 
> temporarily stop, amanda waits for taper to finish some more 
> DLE's, before the dumpers can continue with transferring 
> their data.  Could it be there is a timeout (in the
> firewall) that closes idle connections after some time?
> You should see some indication in the amdump.0 file (or 
> amdump.1 for previous runs etc) about holdingdisk contention.
> 

I had thought of the holding disk issue as well. My total dumpsize
should be 11GB now...my holding disk was at 10GB. I increased it to 20Gb
yesterday, but I still failed this morning. :( 

Let's see if syncing my servers fixes it. I got a good manual level 0
dump today...I'll report in tomorrow morning.

Thanks for your help.