On Mon, May 2, 2011 at 5:29 PM, Lars Ellenberg <lars.ellenb...@linbit.com> wrote: > On Mon, May 02, 2011 at 04:04:56PM +0200, Andrew Beekhof wrote: >> > Still, we may get a spurious failover in this case: >> > >> > reachability: >> > +++++++++_____++++++++++++_____________________________ >> > Node A monitoring intervals: >> > + - + + + - - - - - >> > Node B monitoring intervals: >> > + + - + + - - - - - >> > "dampening" interval: |---------| >> > >> > Note how the "dampening" helps to ignore the first network "glitch". >> > >> > But for the "permanent" network problem, we may get spurious failover: >> >> Then your dampen setting is too short or interval too long :-) > > No. > Regardless of dampen and interval setting. > > Unless both nodes notice the change at the exact same time, > expire their dampen at the exact same time,
This is where you've diverged. Once dampen expires on one node, _all_ nodes write their current value. > and place their updated > values into the CIB at exactly the same time. > > If a "ping node" just dies, then one node will always notice it first. > And regardless of dampen and interval settings, > one will reach the CIB first, and therefor the PE will see the > connectivity change first for only one of the nodes, and only later for > the other (once it noticed, *and* expired its dampen interval, too). > > Show me how you can work around that using dampen or interval settings. > > -- > : Lars Ellenberg > : LINBIT | Your Way to High Availability > : DRBD/HA support and consulting http://www.linbit.com > _______________________________________________ > Linux-HA mailing list > Linux-HA@lists.linux-ha.org > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems > _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
