[Fwd: [openib-general] ifconfig ib0 down and then up vis a vis IP connectivity]

[Hal Rosenstock]

I have confirmed that this is an ARP cache issue on the remote machine.

The remote node is responding to the old DQPN of the machine whose IPoIB
interface was downed and then brought up. When it is brought up again,
it has a different QPN. The remote node still has the old QPN cached
until it times out and sends back to the old QPN which is discarded on
the local node.

It behaves just like a hardware address change for an IP address for
which a remote node had communication with previously (and cached it's
MAC address).

-- Hal

-----Forwarded Message-----

From: Hal Rosenstock <halr at voltaire.com>
To: openib-general at openib.org
Subject: [openib-general] ifconfig ib0 down and then up vis a vis IP connectivity
Date: 02 Nov 2004 14:45:03 -0500

Hi,

What is the ARP timeout in Linux ?

If I down and then up the ib0 interface, there is some delay before
connectivity is restored despite the fact that it is successfully
(re)attached to the multicast groups and that all the QPNs seem to be
the same. After some time period, connectivity is restored. Any idea on
what is different ? It seems like it is an ARP cache issue.

Thanks.

-- Hal

_______________________________________________
openib-general mailing list
openib-general at openib.org
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general