Again, I am not a dev, but my understanding was always that one ring or the other was used, not both. When I experimented with RRP, I saw that the rings would fail over and recover, implying to me that no, they don't use both at the same time.
digimer On 17/10/13 10:37, Moullé Alain wrote: > Hi > About "switching between rings" , the information I had was that it is > dependant to rrp mode, > and in the case rrp mode is active, the information I had was that both > rings were used at > the same time ... > It is it right or wrong ? > Thanks > Alain > > Le 17/10/2013 16:25, Digimer a écrit : >> On 17/10/13 03:08, Ulrich Windl wrote: >>> Hi! >>> >>> Nice explanation! I have one question: With multiple rings, is Corosync >>> expecting the tokens to rotate with the same speed? I'm thinking of a >>> scenario >>> where both rings operate with different speeds, so the token will >>> rotate at the >>> same speed at low or medium network load, but might rotate with >>> different >>> speeds when the slower ring uses the full bandwidth. I have the >>> impression that >>> Corosync makes one of the rings as faulty (for less than one second), >>> then. >>> >>> Regards, >>> Ulrich >> I am not a dev, so I might not understand your question properly. >> However, as I understand it, corosync uses one ring or the other. So if >> a ring is considered faulty, it switches over to the other ring. How >> this is determined, I do not know, though I suspect it's similar to how >> node failure occurs. >> >> As for speed, tokens are passed around as fast as possible. If the >> network is slow for some reason, it will still go as fast as possible, >> it just won't be as fast as the other ring. If the speed drops too much, >> corosync will think a node has failed when it has not, which is why low >> latency networks are needed for corosync (or you adjust the timing >> values to be long enough to account for the slower speeds). >> >> I'd love to hear a corosync dev chime in on this. >> > > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
