Stop optimizing call and see if that resolves the problem. Also how are you indexing? (point 3 above). Are you using CloudSolrClient or manually sending requests to any node?
Thanks, Susheel On Tue, Aug 22, 2017 at 9:27 AM, Shreya Kampli <shr...@bigbasket.com> wrote: > Hi, > > I have setup a solrcloud with 1 shard and 3 replicas on solr 6.6. > Every index operation makes an explicit commit so that the documents are > available immediately. The indexing happens every 5 minutes for few 100's > of documents. > > My problem is that I see that the replica nodes are frequently going down > but the leader is always up. The logs show that it is receiving replication > requests. > There is no issue with heap memory or CPU load. > > My questions are: > 1. How do I find the reason for my nodes going down? > 2. Do the replicas go down while syncing from leader? > 3. I pass indexing requests to any node , does it forward the request to > the node shown as leader in SOLR admin console? If so, why do I see tlog > data in all the other nodes. > 4. I also run optimize call every hour to merge the index on only one node, > is this one of the reason for node going down? > > Can somebody please help me to figure this out? > > > -- > Regards, > Shreya >
