"Stefan G. Weichinger" <monitor AT oops.co DOT at> writes:

> Hi, Kevin,
>
> on Freitag, 29. Oktober 2004 at 06:42 you wrote to amanda-users:
>
> KD> I figured out the problem, after modifying "amadmin disklist" so that
> KD> it prints out the spindle number, at least in amanda-24.4p3.
>
> KD> I had lines in disklist which looked something like this:
>
> KD> localhost //bunting/c$ root-sambatar eth0
>
> KD> eth0 was parsed as a spindle number of 0, in addition to being parsed as
> KD> an interface value of eth0.
>
> Errm, I could have seen this ... was late here ...
>
> KD> It looks like maxdumps is picked up correctly, but the spindle value
> KD> of 0 may be preventing multiple dumpers from running.  I'll give it
> KD> another try soon, and suggest a few patches to the amanda-hackers list.
>
> The file format for disklist is:
>
> hostname diskdev dumptype [spindle [interface]]
>
> which means that you have to define a spindle if you want to define an
> interface. So I don't think this behavior is a bug.

There is a bug in the parsing.

If eth0 is interpreted as the spindle, then why does the interface get
interpreted as ETH0?

    host localhost:
        interface ETH0
    disk //bunting/c$:
         ...
        spindle 0

Also, the parser could notice that eth0 is not an integer.  Using
strtol rather than atoi in diskfile.c will allow checking for errors
in parsing.  

Essentially, atoi interprets "eth0" as 0, but skip_integer doesn't
skip over anything, so eth0 then gets interpreted as "eth0" for an
interface.

Did interface get introduced after the spindle number?  The CVS
interface isn't working now.

-- 
Kevin Dalley
kevin AT kelphead DOT org