Finally solved the problems

1 - The amanda host had a secondary IP alias on the LAN NIC, and Amanda was
swapping between the two IPs for its source address. Thats fine for clients
on the LAN, but the firewall didnt like that much...

2 - Once that had been fixed, I found a nasty `bug` in my firewall software
(dont ask!) that I`ll be chasing up my support people about.

Anyway, its working fine now....

Steve

----- Original Message -----
From: "Gene Heskett" <gene_heskett AT iolinc DOT net>
To: "Steve Loughran" <stevelml1 AT scee.sony.co DOT uk>; <amanda-users AT 
amanda DOT org>
Sent: Friday, December 20, 2002 12:20 AM
Subject: Re: index tee cannot write [Connection timed out] ??


> On Thursday 19 December 2002 10:24, Steve Loughran wrote:
> >Hi all
> >
> >I have been to google and egroups to try to find a fix for this
> > but no luck so far...
> >
> >All my internal hosts (i.e. IRIX/Redhat/Solaris hosts connected
> > via LAN) worked fine, first time, every time, but my first client
> > host through a firewall fails during the dumping process.
> >
> >External host is RH 7.3, RPM version on amanda have been removed,
> > and v2.4.3 has been installed from source. I have done exactly
> > the same on this host as the 20-25 other RH boxes I have on the
> > LAN.
> >
> >amcheck reports the that its test worked fine, the firewall shows
> > no dropped or denied packets during the amdump run, the NAT and
> > rules are correct, all the source and destination IPs and ports
> > are as they should be, but I cannot work out why this is
> > happening
> >
> >amstatus shows the estimate phase has completed, and is now
> > waiting for dumped data from the host:
> >
> >host-1:/                     0 5135470k dumping       32k (
> > 0.00%) (14:41:28)
> >
> >The 32K block is the amanda header, and does not contain any data
> > from the client.
> >
> >I have plenty of drive space on the tape server for the log files,
> >permissions are correct on all directories, the index file for
> > this host/drive has been created but just sits there at 0 bytes
> > long.
> >
> >-rw-------    1 amanda   disk 0 Dec 19 14:41
> >index/host-1/_/20021219_0.gz.tmp
> >
> >The client shows this in the syslog:
> >
> >sendbackup[24693]: index tee cannot write [Connection timed out]
> >sendbackup[24691]: error [/bin/gtar got signal 13]
> >
> >and then it just sits there. I am waiting for dtimeout so it moves
> > onto the next drive on that host to see if it does the same
> > thing, but I have a feeling it will.
> >
> >What have a missed? This is driving me nuts....
> >
> >Any help you can give me would be greatly appreciated. Many thanks
> > in advance.
>
> This is off hand, but it sounds like you need more etimeout.  Or the
> firewall itself is timeing out, I think I heard someone mention
> that possibility a while back up the log.
>
> --
> Cheers, Gene
> AMD K6-III@500mhz 320M
> Athlon1600XP@1400mhz  512M
> 99.20% setiathome rank, not too shabby for a WV hillbilly
>