[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-atm
Subject:    More CLIP weirdness
From:       "Joseph Gooch" <mrwizard () psu ! edu>
Date:       2001-01-26 21:26:03
[Download RAW message or body]

Oh yeah more info.

Here's one I can't see at the moment.

# grep 136 /proc/net/atm/arp
atm0  SVC LLC 48139 192.168.0.136   (expired, ref 2)

# atmarp -d 192.168.0.136
atmarp: No such file or directory
# atmarp -s 192.168.0.136 0.500
atmarp: Address already in use

That IP doesn't show up in atmarp -a at all.

The only solution to this that I see is bringing atm0 down and back up to
clear the entries.  Perhaps atmarpd should clear the kernel entry even if
it's not in its own table?  Or it could attempt to synchronize itself with
the kernel on a regular interval?

I've seen similar issues with LANE as well, although not as recently.  One
of the hosts that wasn't in the arp table returned after I deleted and
recreated signalling from the endpoint to the switch (on the remote side) in
effect manually destroying the VC.

I don't even know how to track it down, what does it mean if the SVC isn't
in /proc/net/atm/vc but it's still there?  Sounds to me like the kernel
entry was deleted but atmsigd didn't signal the switch to destroy it, or
something.  I don't know enough about uni to know :)  Atmsigd is set to use
uni30, should I try uni31 or uni40?  Or are those still partially working.

Later,
Joe Gooch
Carbon Lehigh Intermediate Unit

[prev in list] [next in list] [prev in thread] [next in thread]