On Fri, 12 Dec 2014 11:32:31 +0530 Atin Mukherjee <amukherj@xxxxxxxxxx> wrote: > Issue is cracked now!! > > http://review.gluster.org/#/c/9269/ should solve it. Commit message is > self explanatory, however just briefing the problem and solution: > > If we look at the mgmt_v3-locks.t, it tries to perform multiple syncop > transactions parallely. Now since these operations are attempted on > two different volumes they are allowed. The current implementation > uses xaction_peers to store the list of peers which are part of the > cluster before the syncop transaction begins, this is just to ensure > the staticness of the peer list even though new peer is been added in > between. However this xaction_peers was stored as global which means > multiple transaction was manipulating the same list which might end up > having a corrupted xaction_peers list. > > Please note the mentioned patch only fixes the syncop framework, I've > also identified the same problem for op-sm as well which will be > addressed in a different patch. Awesome. :D It'd be really good to get to no spurious regressions happening, ever. :) + Justin -- GlusterFS - http://www.gluster.org An open source, distributed file system scaling to several petabytes, and handling thousands of clients. My personal twitter: twitter.com/realjustinclift _______________________________________________ Gluster-devel mailing list Gluster-devel@xxxxxxxxxxx http://supercolony.gluster.org/mailman/listinfo/gluster-devel
