-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Use channel bonding...
http://www.mjmwired.net/kernel/Documentation/networking/bonding.txt - -- Flávio do Carmo Júnior aka waKKu -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (MingW32) Comment: http://getfiregpg.org iEYEARECAAYFAklBwRAACgkQgyuXjr6dyku0tQCeOftcgUfJAbnnGeOZu25VA52p obcAn2egYD9SLjXbHnnNtRlxKRsfXKU8 =3Eg7 -----END PGP SIGNATURE----- On Thu, Dec 11, 2008 at 11:32 PM, Michael O'Sullivan < [email protected]> wrote: > Hi everyone, > > I have been trying to set up a 2-node cluster where both nodes have > multiple links to the cluster. Initially, I specified two IP addresses for > each cluster node in /etc/hosts, but this seemed to confuse the fencing and > make the cluster hang. > > Now I am trying to use a link equalizer (TEQL device, see the Linux > Advanced Routing and Traffic Control - lartc.org for details) which uses a > single IP address and shares the load across the two links. However, some > packets may be delivered out of order... > > Unfortunately, using the TEQL device makes the 2-node cluster have a > split-brain problem that I can't seem to remedy. I changed /etc/hosts to use > one of the actual NICs instead of the TEQL device and got the cluster going, > but as soon as I go back to the TEQL device node 1 thinks node 2 is offline > and node 2 thinks node 1 is offline... > > Does anyone have any suggestions to fix this? Or alternatives for using a > single IP address? (Note, I have not used ethernet bonding as it does not > utilize the full capacity of both links whereas I think TEQL will.) > > I'm happy to supply any extra information required. Thanks in advance for > any help, Mike > > -- > Linux-cluster mailing list > [email protected] > https://www.redhat.com/mailman/listinfo/linux-cluster >
-- Linux-cluster mailing list [email protected] https://www.redhat.com/mailman/listinfo/linux-cluster
