Hello, in our setup we have two Solr servers and we index documents to both (we keep track of the last sync for each server so they can also be not aligned in a certain moment for example for maintenance). The clients use SolrJ and we have an adapter that creates “sessions” using round robin. Once a client has a session, it’s tied to one server (for consistency). If the server goes down, the session is switched to the other server and the same query is repeated. If both servers are down the adapter returns an error.
This setup can be easily extended for any number of servers. We find this setup very useful to do maintenance or upgrading the Solr version. On 2022/03/13 15:23:10 Sam Lee wrote: > How do I run Apache Solr on two servers such that I will still be able > to index and query even if one of the servers is taken offline? In other > words, I am looking for high availability with automatic failover. > > I have looked at SolrCloud. However, it uses Apache ZooKeeper which > requires at least 3 servers to be able to tolerate the failure of 1 > server. Are there alternative methods? > > I only have two servers that are able to run Solr. Supposing I have a > third low-spec server, is it a good idea to run Solr (SolrCloud) on only > two of the three servers, while running ZooKeeper on all three servers? > However, this looks like a waste because the third server would > essentially be doing nothing. Is there a better method with standalone > Solr? > — Ing. Andrea Vettori Sistemi Informativi B2BIres s.r.l.
