On Wed, Jan 12, 2011 at 10:15 PM, Ravi Pinjala <pstatic@xxxxxxxxx> wrote: > Quick question about cluster upgrades. Are the Ceph protocols nailed > down enough that it's safe to upgrade a cluster incrementally and mix > versions of the software, or do I have to take the cluster offline and > upgrade everything at once? Interoperability between versions is something we try to maintain, but at this point isn't guaranteed. I can't think of any situation we've had where trying to run mixed versions would break your on-disk data, though, it just might crash your older daemons. Perhaps somebody else will have other ideas, but if I were upgrading a cluster I think I'd leave everything running, then upgrade one type of daemon at a time. ie, install the new software on the monitor machines, then restart the monitor daemons so they're using the new code, then install on the OSDs, then install on the MDSes. -Greg -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
