On 06/10/13 01:07, Wolfgang Hennerbichler wrote: > On 10.06.2013, at 09:48, James <purpleidea at gmail.com> wrote: > >> On Mon, Jun 10, 2013 at 1:16 AM, Wolfgang Hennerbichler <wogri at wogri.com> wrote: >>> On Jun 9, 2013, at 23:30 , Mohit Anchlia <mohitanchlia at gmail.com> wrote: >>> >>>> volume changes should survive the node failures >>> >>> even if ALL nodes fail? >> I mean if all the nodes "fail", then this is equivalent to setting all >> of the hardware on fire... > > Hah :) > Failing in my concern was meant to be a power loss on all nodes. Assume two nodes on the same site, suffering from a power outage. > >> So no, it won't persist that. But each individual node could fail, and >> you could rebuild the cluster accordingly (assuming you are using >> replication...) > > Sure. > >> Think about it... Hope this helps. > > Sorry to be insistent, but this really bothers me. It is persisted to disk. You can check this yourself by going through /var/lib/glusterd and looking at the .vol file and others. -- Mr. Flibble King of the Potato People
