On 7/8/2024 03:25, Roberto Maggi @ Debian wrote:
I'm totally new to apache ecosystem and, after studying solr9
documentation I can't clarify a doubt.
Once I created the zookeeper cluster as well as the solr one, the latter
cluster elects the "leader" and goes online.
At this point I'm wondering what will the endpoint be and how can I
offer the "clients" a stable endpoint toward the time-by-time leader?
Solr has no built-in mechanism to provide a stable endpoint for a whole
cluster. You need something like a load blaancer to create an endpoint
that will always work regardless of which servers in the cluster are up.
As long as enough servers in the cluster are up so that the whole index
is available, you can send requests to any machine in the cluster and
they will be directed to the correct machines to handle the request.
If you need to be able to send updates directly to leaders, then you
need a client that is cloud-aware. The Java client named SolrJ is
cloud-aware. I believe at least one of the Python clients (third-party)
is also cloud-aware. If only cloud-aware clients are being used, then
you do not need a load balancer, as such a client always knows the state
of the whole cluster.
Thanks,
Shawn
