Just thought I would put 'some' closure on this. The problem I was having
was down to a dodgy RAID array, the backups now finish within 7 hours as
opposed to 15. However, I reckon I could get them down to 5 hours if I
could get the inparallel option working properly.
I had a reply from a poster saying that  "That is what Amanda does.  Even
if it is doing a level 0 backup, it will call gtar that way. I solved this
by using calcsize instead of gtar for my estimation phase."
I use a wrapper to get the sizes which all happen very fast, this
listed-incremental thing happens during the actual backup. Running the
command by hand takes a while, stripping out the listed-incremental flag
speeds it up no-end. There must be a way to stop it from doing this
command. Failing that, I will have to work out how to filter this flag out
from being passed onto gtar.

|---------+---------------------------->
|         |           Jason Walton     |
|         |                            |
|         |           25/07/2003 15:08 |
|---------+---------------------------->
  
>-----------------------------------------------------------------------------------------------------------------------|
  |                                                                             
                                          |
  |       To:      amanda-users AT amanda DOT org                               
                                                 |
  |       cc:                                                                   
                                          |
  |       Subject: Re: amanda inparallel not working on large filesystems 
(Document link: Jason Walton)                   |
  
>-----------------------------------------------------------------------------------------------------------------------|



Hello,
I have left this alone for a month now hoping all the various suggestions I
received might help. Nothing did.
Over the previous month, I have moved the backup onto seperate network
cards, seperate networks, specifed spindles, increased the holding disc
space, reduced the holding disc space, increased max dumps etc, reduced max
dumps etc.
Nothing appears to work. From my observations, the gtar on the remote
machine is using "listed_incremental" even though I am specifying no
incrementals to be performed. I don;t know if this is important or not, but
if I run the command by hand it takes forever to run, if I omit the
list-incremental option on the command line. gtar flies through. Note that
this may be a totally seperate issue to the problem I am facing.
The backups still operate synchronously, they both start roughly the same
time, but then one waits for the other to finish before it wakes up and
continues it's backup (this can mean it lays dormant for about six hours).