Skip to Content
Jan 30, 2019 at 03:32 PM

SolrServerConfig.maxTotalConnections vs tomcat.maxthreads


Hello experts, experts in the making, et. al.,

We are tuning up performance for Solr. We have a cluster with a master and a couple of slaves. Question for you is, is there a rule of thumb to follow when it comes to setting the values for:

  1. SolrServerConfig.maxTotalConnections

  2. SolrServerConfig.maxTotalConnectionsPerHostConfig

Our understanding is that, this would act like a pool to limit the number of HTTP requests coming from hybris that would execute a query on the solr nodes, is that correct? We have not found any document explaining anything related to Solr-performance. We are using the rule of thumb that, since the platform comes with: tomcat.maxthreads=200 and db.pool.maxActive=90 and it comes with:

 <attribute qualifier="maxTotalConnections" type="java.lang.Integer">
 <modifiers optional="true" />
 <persistence type="property" />
 <description>Sets the maximum total connections allowed</description>
 <attribute qualifier="maxTotalConnectionsPerHostConfig" type="java.lang.Integer">
 <modifiers optional="true" />
 <persistence type="property" />
 <description>Sets the maximum number of connections to be used for the given host config</description>

Questions are: - 200 HTTP requests for 100 maxTotalConnections. If the number of HTTP requests grows (tomcat.maxthreads), would you keep that proportion? - Does maxTotalConnections set the total number of connections for the cluster? Is that controlled by jetty?

Kind regards,