On 03/17/2012 03:07 PM, Kiss Bence wrote: > Hi All, > > I still have the issue with corosync 1.4.2 on fedora-14. I disabled > secauth turn on debug and add a third node standing-by for the quorum. > > Can someone tell me what's wrong? You can see debug logs of corosync > from the three nodes, config and version infos also in the tarball here: > > http://noc.elte.hu/~bence/public/corosync-142.logs.tar.bz2 > > virt1 and virt2 for services virt3 in stand-by for quorum. They are on > the same real machine with 4G ram and 4 core AMD processor. Each virtual > machine has 1 cpu core and 650M ram. > > At Mar 17 11:24:26 on virt2 corosync exited cleanly on system reboot. At > 11:24:55 it started after the reboot. > > On virt1 the corosync gain knowledge of new processor virt2 only at > 11:29:16 on virt3 it is member only at 11:29:17. > > Thanks, > Bence > _______________________________________________ > discuss mailing list > discuss@xxxxxxxxxxxx > http://lists.corosync.org/mailman/listinfo/discuss Bence, Please run corosync-blackbox after killing the node. I expect some problem with multicast. Is STP enabled on your virbr? If so, you might try disabling it. The fastest way to get to a working solution is probably to just use udpu. See man pages. Regards -steve _______________________________________________ discuss mailing list discuss@xxxxxxxxxxxx http://lists.corosync.org/mailman/listinfo/discuss
