Hmmm, I don't think anyone's really documented this as the supposition is that one would only run embedded for sandboxes and set up an external ensemble "for real".
So, with the caveat that I haven't personally tried this, I'd add external zookeepers as part of an ensemble that contained my embedded zookeepers. That means your ZK configurations pointing to your embedded ZK instances. At that point the external ZK _should_ replicate the data from the embedded instances to the external ones. Then shut down all your Solrs and change your external ensemble configurations to only point to each other (i.e. take the embedded stuff out). Now start your solrs pointing to the external ensemble. As I said, though, I haven't personally done this so go ahead and give it a try. Or, if you're feeling _really_ brave, just copy the zookeeper data directory from the place the embedded Zookeeper put it to the place you specify in your external ensemble then start your external ensemble. Theoretically, this should work bu tas I said I haven't tried either of these personally. Best, Erick On Tue, Mar 15, 2016 at 9:59 AM, Rachid Bouacheria <willi...@gmail.com> wrote: > Hi, > > I am running solr 4.x on 3 servers with zookeper embedded in prod. > > Each servers has 1 leader and 2 replicas. > > I want to switch zookeper from embedded to standalone. > > I want to know if the steps are documented anywhere? I could not find them. > > I am worried my index will get messed up if in the transition. > > Thank you very much!
