Hi,
your assumption about the hostnames is correct.
What you are doing is one of the big NO!NO!'s of amanda.
Never list a single host in multiple concurent amanda runs, or,
in your case, with different names in a single amanda configuration.
amandad can only handle one request at the same time.
Only solution is to use only one hostname per physical host in disklist.
No other way out there.
Christoph
Toralf Lund schrieb:
> For the "hostname" parameter of my disklist file, I'm normally using
> various hostname aliases (defined in /etc/hosts or DNS) based on the
> function of the host, rather than the official host name. Different
> entries may use different aliases for the same host. For instance, I have
>
> amanda-server /dumps/amanda/ks { full-tar [ ... the usual special
> options for dump dir ]
>
> fileserv /u full-tar
>
> where "amanda-server" and "fileserv" are actually the same host (now,
> but that may change in the future, which is the whole argument behind
> the aliase-based setup.)
>
> I have never noticed any problems with this, but I'm now reviewing the
> whole backup process (as you may have gathered), looking closely at the
> logs etc., and it occurs to me that I'm getting a lot of
>
> error result for host fileserv disk /u: Request to fileserv timed out
>
> or similar even though I *know* the host is up and has amandad enabled,
> and the entries are successfully backed up on the next run or so.
>
> Furthermore, it looks like this is related to the alias-based setup; it
> seems like I often will get timeout message for "amanda-server" entries
> during runds where the "fileserv" ones are handled correctly, and vice
> versa. Is this a correct assumption? Why does it happen? Is there a way
> around it (obviously, I can change the hostname in disklist, but apart
> from that)?
>
