[Pacemaker] Redundant rings vs one bond based ring

Steven Dake sdake at redhat.com
Tue May 18 17:41:11 EDT 2010


On Tue, 2010-05-18 at 23:16 +0200, Gianluca Cecchi wrote:
> Hello,
> based on pacemaker 1.0.8 + corosync 1.2.2, having two network
> interfaces to dedicate to cluster communication, what is better/safer
> at this moment:
> 
bonding

> 
> a) only one corosync ring on top of a bond interface
> b) two different rings, each one associated with one interface
> ?
> 
> 
> Question based also on corosync roadmap document, containing this
> goal:
> Improved redundant ring support:
> The redundant ring support in corosync needs more testing, especially
> around boundary areas such as 0x7FFFFFFF seqids.
> Redundant ring should have an automatic way to recover from failures
> by periodically checking the link and instituting a recovery of the
> ring.
> 
> 
> BTW: if a link fail, what is the current "manual" command to notify
> the CCE when it becomes available again? 

corosync-cfgtool -r

> 
> 
> Thanks,
> Gianluca 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf





More information about the Pacemaker mailing list