Replication in Solr Distributed Configuration

Question

I am trying to add replication for every subordinate in Solr cloud to make the whole cluster more reliable. The idea is to distribute one fragment to several machines in order to work in a cluster, even if one of the machines is turned off.

Solr has an excellent built-in replication feature that works great in a regular (non-distributed) Solr setup. However, when I add this function to the Solr cloud, I get an error saying that the /solr/replication handler was not found (and in fact it does not appear in the Solr output when Solr starts, although it is defined in solrconfig.xml ).

SolrWiki also points to a distributed update handler , but from the author’s note, I understand that it does not actually replicate the data, but simply sends new documents to one of the nodes defined in config.

My questions:

• Why doesn't Java-based inline replication work in Solr distributed setup? Is it a Solr cloud problem, a more fundamental problem, or just an error in my configuration (I checked everything several times, but still could have missed something)?
• What are other good replication approaches in distributed Solr?

Answer 1

Although I'm a bit late for the party here on the old question; the answer is that Solr Cloud handles replication internally. Page The Solr Cloud widget page explains this in detail. If you set numShards = 2 and add more servers (so that you only have four), the splinter will be replicated to the new servers - make sure that your splinter lives on more than one node.

To answer your questions directly;

• SolrCloud does the replication setup and logic for you, and you should let it do it yourself, instead of introducing “manual” replication of the settings in the mix.

• The whole point of SolrCloud is to hide the logic of replication and sharing, allowing you to simply add more servers as they become available. Of course, you can create logically separate cloud clusters and indexes for both, but this is not necessary.