Thanks for the suggestion. I have made solr host as the pod dns . Solr operator is great but needs lots of in depth of k8 expertise. I did a helm chart provided by lucidworks. Couple of quick queries what I am observing..
Whenever my pod comes up , I create a collection And then I use ADDREPLiCA. Collection creating takes node set as empty , max shards on a node as 1. So what I have observed is that ADD replica commands fails most of the time with 500 error On Mon, 4 May 2020 at 2:44 AM, Jan Høydahl <jan....@cominvent.com> wrote: > Solr is built to assume nodes/pods with stable host names and indices. > Please consider deploying solr with > https://github.com/bloomberg/solr-operator which takes care of this for > you. > > Jan Høydahl > > > 3. mai 2020 kl. 19:25 skrev Vishal Vaibhav <defvis...@gmail.com>: > > > > > > > > Hi I could finally deploy solr cloud 8 using zk ensemble on Kubernetes. > > However I am seeing something strange. > > > > The first thing since k8 keeps rescheduling the pods, the node list gets > > filled up like this even the node are dead . Also I am using solr > exporter, > > so solr exporter keeps trying to hit the dead replicas . Can I get rid of > > them somehow . > > <IMG_4637.jpg> >
