On 10/25/2011 08:11 AM, Yunkai Zhang wrote: > Hi Steven Dake: > > In our team, my coworkers and me are developing sheepdog which use CPG > service provided by corosync. > > We found that sheepdog will receive more than one confchg msg when > network partition occur. For example, suppose the cluster has 4 > nodes: N1, N2, N3, N4, and they form a single-ring initially. After a > while, network partition occur, the single-ring divide into two > sub-ring: ring(N1, N2, N3) and ring(N4). The sheepdog in the ring(N4) > will receive the following confchg messages in turn: > Memb: N2,N3,N4 Left:N1, Joined:null > memb: N3,N4 Left:N1,N2 Joined:null > memb: N4 Left:N1,N2,N3 Joined:null > Assuming there are no messages sent by cpg between these leaving events, this looks like a bug if totem generates 1 configuration change on this partition. I've copied Angus since he wrote the latest incarnation of that code ;) Regards -steve > I have read the corosync's code, and found that corosync send these > confchg messages in function: downlist_master_choose_and_send (in the > for loop). > > Why we send so many confchg message to CPG client(here is sheepdog) ? > The fist two message will make client confusing and do more > meaningless work. > > Can we give the user a selection to enable/disable this downlist > messages? Sheepdog just want to receive the last msg in this case. > > Thank you! _______________________________________________ discuss mailing list discuss@xxxxxxxxxxxx http://lists.corosync.org/mailman/listinfo/discuss
