On Fri, Sep 25, 2009 at 04:55, Jan Friesse <jfriesse@xxxxxxxxxx> wrote: > Edson, > so as I expect, you have UPS not PDU. Problem with that UPS is only one > (respectively two) communication connectors. Because EVERY node in > cluster must be able to fence EVERY other node and it's not possible to > achive this with 1 connector (ok, theoretically it is possible to have 2 > cheap UPS, cross connected for 2 node cluster, but we don't support this) Yes, that's exactly what I have here. > Maybe that card helps, but I'm really not sure, about support in fence > agent. In case you will decide to buy it, and you will be able to turn > on/off plugs (you must be able to turn on/off only ONE plug, all of them > is useless) by web and SNMP, you can try fence_apc_snmp. In case it will > not work, you can try to send there SNMP dump and I should be able to > implement support to fence_apc_snmp. I thank you very much, Jan, but I decided to keep with only a single fencing method. I think it will be enough to me, anyway. (My original intention on using nobreak fencing methods was to avoid cluster to stop due to power cables disconnection on a node, or something else that cuts off its power completely, what would make iLO interface unreachable, and the other node won't start the services because it can't be sure that the other is down. But I realized that it doesn't make sense, just because it is very unlikely to happen.) > Regards, > Honza -- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
