There isn't nearly enough information to say much of anything here. Saying "Poi" makes me wonder if you're using the extracting request handler, in which case I'd recommend you move it off Solr, see: https://lucidworks.com/2012/02/14/indexing-with-solrj/
you might review: https://wiki.apache.org/solr/UsingMailingLists Best, Erick On Tue, Jul 18, 2017 at 9:08 AM, Satya Marivada <satya.chaita...@gmail.com> wrote: > Hi All, > > We are using solr-6.3.0 with external zookeeper. Setup is as below. Poi is > the collection which is big about 20G with each shard at 10G. Each jvm is > having 3G and the vms have 70G of RAM. The processors are at 6. > > The cpu utilization when running queries is reaching more than 100%. Any > suggestions. Should I increase the number of cpus on each vm. Is it true > that a shard can be searched by a cpu and not multiple cpus can be put to > work on the same shard as it does not support multiple threading? > > Or should the shard be split further? Each poi shard now is at 10G with > about 8 million documents. > > > [image: image.png] >
