This one time, at band camp, Gene Heskett wrote:
>On Tuesday 15 February 2005 19:49, Jamie Wilkinson wrote:
>>This one time, at band camp, Gene Heskett wrote:
>>>>If you have plenty of holdingdisk space, I would think they would
>>>> go there. Even with the default reserve of 100%, you should still
>>>> get the incremental. You might have to look in the debug files to
>>>> see why they were completely skipped.
>>>
>>>If the dumps are too big, and the holding disks reserve is at the
>>>default of 100%, the dump *will not be done*.
>>>
>>>Set the reserve on the holding disk, if you've got a ton, to
>>> something like 30%, which will let it do fulls till there is only
>>> 30% of the holding disk left.
>>
>>The reserve is set to 40%.  Amanda is not dumping anything once the
>> planner decides it won't fit on the tape, according to the debug
>> log.
>
>Whats 60% of the holding disk in gigabytes?
>The tape is how many gigabytes?
>The dle is how many gigabytes?

The holding disk has 646GB available, so approximately 387G.
The tape is 100GB.
The DLE is usually 86GB, the partition in question is currently 90GB.

It's probably worth mentioning that the server is 2.4.4p1 from RHEL 3ES U4,
because we seem to have dissonance between the behaviour I'm seeing and the
behaviour you're describing.

To reiterate, Amanda does complete the estimate for all DLEs.  The large DLE
gets started and some smaller ones that will fit on the tape after it also
start, but the planner says no more.

What I'm after is a way to tell amanda that I understand it won't be able to
fit any more, but dump the other incrementals anyway.

Since re-reading the manpage, thanks to some advice here, I've increased
runtapes to 2, which I think will solve this.  I didn't realise that it was
an upper limit rather than an absolute, although I'm still unconvinced that
Amanda will continue to use only one tape except in this extreme case; we'll
see in the next week how well it goes.

Thanks for all your replies.