Re: Linux Routing Issue
From: prg (rdgentry1_at_cablelynx.com)
Date: 30 Dec 2004 20:27:29 -0800
Gus Zernial wrote:
> (The below all relates to attempts to set up open source software
> called JavaHMO, on a Linux box, to publish content to a Tivo box
> on my home network).
> I have a home subnet 192.168.0. inside a Linksys firewall router.
> On the subnet are wired Windows PC's and one wired Linux box. Also
> wired to the firewall router is a wireless hub, and communicating
> with that hub is a Tivo box with a USB wireless card. (Tivo is a
> media recorder which runs Linux and has an IP address).
> The Windows PC's can publish content to the Tivo box, but the Linux
> cannot do so. Network diagnostics are:
Not sure what you mean by "publish content" but doubt (for now) that
> 1) The Windows PC's and the Linux box can all ping each other and
> everything on the subnet, including the firewall router
> the wireless hub (192.168.0.50) and the Tivo box (192.168.0.51).
Also may not matter, but it's always nice to know the precise
brand/model# of network gear -- your Linksys and wireless "hub" viz.
That way we can download a manual for the thing (maybe).
> 2) On a Windows PC, I can tracert to everything on the subnet,
> the firewall router (192.168.0.1), the wireless hub (192.168.0.50)
> the Tivo box (192.168.0.51). All tracerts are direct - no
Direct, ie., not routed. As expected.
> 3) On the Linux box, I can traceroute to the Windows PC's, and the
> wireless hub (192.168.0.50). These traceroutes are direct - no
> intermediate hops.
> On the Linux box I *cannot* traceroute to the
> firewall router (192.168.0.1), or to the Tivo box (192.168.0.51).
Ie., you get no return packets (probably). The only way to know for
sure is to fire up tcpdump or, better yet, Ethereal and sniff the wire.
Ethereal is easy to use and you can watch it in "real time" on screen
as well as save to capture file. Either or both devices could filter
this particular traffic.
> 4) The Kernel IP routing table on the Linux box is:
> Destination Gateway Genmask Flags Metric Ref
> 192.168.0.0 * 255.255.255.0 U 0 0
> 169.254.0.0 * 255.255.0.0 U 0 0
> default 192.168.0.1 0.0.0.0 UG 0 0
> I'm assuming that the fact that the Linux box cannot publish content
> the Tivo box is reflective of the fact that the Linux box cannot
> traceroute to the Tivo box.
Try traceroute -I to make Linux traceroute behave like Win tracert.
Traceroute uses UDP by default while Win tracert uses ICMP -- as the
others have noted.
The Win (XP?) boxes are getting through. You can use a Win version of
Ethereal to capture some "good" session connection packets. What have
they got that the Linux box don't?
> I'm stumped on what's happening on the
> network - I always thought if you could ping something you could
> a route to it.
Not so, Grasshopper. Any number of reasons why either one of these
might fail and the other succeed. In fact, the failure of _both_ is
not conclusive :(
> Anyway, I'm hoping some of you network experts out there
> can tell me what's wrong and how to fix!
At this point, you have done the first order of business and reported
it with unusual calm for someone in the throws of "net failure". Don't
That you can ping is _good_ news.
Try traceroute -I from Linux and report results.
$man traceroute and look at the -p option and try another port# (can't
Get Ethereal (probably already have it for Linux but might as well get
the latest) for Win here (follow directions re: WinPcap):
It's easy to install, start capture, tell it to update the screen
while capturing, then stop capture when you've collected the
"interesting" stuff. There is a way to post a text version if we ask.
Would be nice if you could connect a PC to the wireless device and
sniff from "the other side".
Also double check your JavaHMO setup on Linux -- _lots_ of dependencies
whose presence will vary with distro/version. Tell us your Linux
flavor (always needed). At this point we have no reason to believe
that JavaHMO is connecting/sending anything to Tivo. Sniffer will
confirm this. Check the JavaHMO logs -- turn on debugging.
Right now, I would not put too much meaning in the failed traceroute.
Sniff the wire, double check your JavaHMO setup, and turn on debugging
in its logs.
Don't forget to tell us the model #s of the Linksys and the wireless
email above disabled