Thanks. Yeah we typically restart the nodes in the minor version to force resync.
But is there a problem with letting schema disagreement running for a long time? Thanks. John On Mon, Jul 6, 2015 at 2:29 PM, Robert Coli <[email protected]> wrote: > > > On Thu, Jul 2, 2015 at 9:31 PM, John Wong <[email protected]> wrote: > >> Hi Graham. Thanks. We are still running on 1.2.16, but we do plan to >> upgrade in the near future. The load on the cluster at the time was very >> very low. All nodes were responsive, except nothing was show up in the logs >> after certain time, which led me to believe something happened internal, >> although that was a poor wild guess. >> >> But is it safe to be okay with schema disagreement? I worry about data >> consistency if I let it sit too long. >> > > In general one shouldn't run with schema disagreement persistently. > > I've seen schema desynch issues on 1.2.x, in general restarting some > unclear subset of the affected daemons made them synch. > > =Rob > >
