Search code examples
solrdatastax-enterprisedatastax-startup

dse enterprise solr re-indexing


Is there a way to re-index a solr core without impacting applications that rely on that core? For example, can we spin up a new replacement core and let it get indexed fully before swapping out and decommissioning the old core?

In our use case, we cannot afford to have partial data available to our applications - which is what will happen if we do an in-place re-index. Currently, it takes anywhere between 24 - 36 hours to fully re-index our core.


Solution

  • If the relevant keyspace is configured with a replication factor of 2 or more, you should be able to do a rolling re-index of your cluster without affecting availability. (i.e. You should be able to use dsetool reload_core <your core name> distributed=false reindex=true.) While a node is re-indexing, it will not service queries for the token ranges it owns, unless there are no other replicas available.