Hi Jiaju, I updated the branch. I fixed two issues when the instance number is implemented.
1. Regardless of whether learner have a quorum, learner fail to renew a ticket when acceptor can't receive a proposer message. 2. booth can't recover the instance number when booth was fail-over in local. I explain first issue. In terms of Paxos, this situation should commit because learner is independent of acceptor. It would appear that a proposer message is single-failure point. So, I fixed a part of acceptor. we move the making of acceptor timer from acceptor to learner. when learner commit, learner make certainly the new acceptor timer. I explain second issue. the current instance number is lost because there's no way to store the instance number. There are some may to resolve this issue. And, I resolved by the following way. the first "renew" run full-round paxos after booth is fail-over. From the second time, booth run the normal "renew". booth can get the correct round after booth is fail-over. If booth boost a round, then the instance number is reset. So, proposer run promise after booth is fail-over. Sincerely, Yuichi 2013/2/5 Jiaju Zhang <jjzh...@suse.de>: > Hi Yuichi, > > On Tue, 2013-02-05 at 14:46 +0900, Yuichi SEINO wrote: >> Hi Jiaju, >> >> I found some issues. So, I want to fix those issues. And, I want to >> replace this patch. >> Please wait to merge this patch. > > OK;) > > Thanks, > Jiaju > > -- Yuichi SEINO METROSYSTEMS CORPORATION E-mail:seino.clust...@gmail.com _______________________________________________ Pacemaker mailing list: Pacemaker@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 Bugs: http://bugs.clusterlabs.org
