hi,
thanks for your help on the topic, but I do not understand what is
the second approach...
could you be more specific on it ???
any example or reference please ??
briner
>
> BRINER Cedric <[EMAIL PROTECTED]> writes:
>
> > the disklist in now composed of more than 100 entries.
> > but the problem reside in the fact that I do have some
> > big server that has about 20 entries in the disklist.
> >
> > So as I read...amanda is not launching more than one backup per host..
> > and during the last amdump amanda do planned to backup 7 partition in
> > the same time...so instead of having a backup that keep for 5 hours it
> > suddenly jump to 11 hours..
> >
> > So my question is :
> > Is there a way to also balance the number of dump level 0 per host ???
> > like a parameter saying that it shouldn't be more than 4 (for example)
> > backup 0 on this host ???
>
> Cedric,
>
> Have you read about the "spindle" parameter in the disk list file? It
> is used to allow parallel dumps only if the disk list entries are on
> physically separated disks. If software compression or network are no
> bottle neck it'll speed up the backup if "max dumpers" is set to at
> least the maximum spindle number.
>
> A better aproach would is to tweak the "small dumpers" and "big
> dumpers" in the source. But that's not host based. There was a
> discussion about it some days ago.
>
> Johannes Niess
