Ok, this was a weird one. It turned out to be due to tcpwrappers. I just
saw a recent post suggesting it. It's odd that inetd closed the port, but
I would guess that that might be caused because the protocol is udp
instead of tcp?
Put this in a FAQ somewhere!
-doug
On Thu, 7 Jun 2001, Doug Silver wrote:
> Ok, now I'm stumped. I have two FBSD 4.3 machines, one inside and one
> outside a firewall. I compiled amanda2.4.2p2 on the internal machine and
> it checks out fine. So, I tarred up libexec, lib/liba*, and sbin/am*, put
> it on the external machine, HUP'd inetd and poof, it still doesn't
> work. It's not a shared library issue because amanda isn't using shared
> libs on FBSD. In fact, I just double checked and the inetd binaries are
> exactly the same (md5 confirmed).
>
> Here are some gory tcpdump details:
> Internal client
> 10:18:45.151846 fbsd43.internal.amanda > amanda-server.internal.902: udp 360
> 10:18:45.152218 amanda-server.internal.902 > fbsd43.internal.amanda: udp 49
>
> External client:
> 10:23:50.461673 firewall.921 > fbsd43.external.amanda: udp 155
>
> /var/log/messages two seconds later reports
> Jun 7 10:23:52 fbsd43 inetd[152]: amanda/udp server failing (looping), service
>terminated
>
> I know the firewall is acting properly because I'm backing up at least four other
>external
> machines, most are FBSD 3.3-4.2.
>
> John - I can run your amandad test script from the command line and it
> gives this:
> # cat amandad.out.74693
> Thu Jun 7 10:34:18 PDT 2001: starting amandad
> Thu Jun 7 10:34:48 PDT 2001: amandad done: status is 1
>
> cat amandad.20010607103418.debug
> amandad: debug 1 pid 74695 ruid 0 euid 0 start time Thu Jun 7 10:34:18 2001
> [snip compile info]
> amandad: dgram_recv: timeout after 30 seconds
> amandad: error receiving message: timeout
> error receiving message: timeout
> amandad: pid 74695 finish time Thu Jun 7 10:34:48 2001
>
> Any other suggestions???
>
> Thanks.
>
> Doug Silver
>
>
>