I could not find any documentation on how the backend deals with unavailable Solr servers in case there are multiple configured in the facet search config, so I tried to take a look at the decompiled code and experiment with it.
To me it seems the default logic is like this:
If a query or indexing operation does not get a result, move on to the next Solr server. The timout for this can be controlled in the backoffice, the default is 5 seconds. See AbstractSolrSearchProvider.
Every 60 seconds check if the Solr server can be reached. If not, remove it from the pool of available servers. This intervall can be changed using the system property solrfacetsearch.solrClientPool.checkInterval. See DefaultSolrClientPool.
If there is only one Solr server remaining, keep it in the pool even if it cannot be reached at the moment.
If a Solr server is removed from the pool, it's gone for good and will return only after a restart when the facet search config is read again.
Is this correct?
Bluesky
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.