Agreed, I do something between 20 and 1000. If the master node is not handling any search traffic, use twice as many client threads as there are CPUs in the node. That should get you close to 100% CPU utilization. One thread will be waiting while a batch is being processed and another thread will be sending the next batch so there is no pause in processing.
wunder Walter Underwood wun...@wunderwood.org http://observer.wunderwood.org/ (my blog) > On Jul 7, 2020, at 6:12 AM, Erick Erickson <erickerick...@gmail.com> wrote: > > As many as you can send before blowing up. > > Really, the question is not answerable. 1K docs? 1G docs? 1 field or 500? > > And I don’t think it’s a good use of time to pursue much. See: > > https://lucidworks.com/post/really-batch-updates-solr-2/ > > If you’re looking at trying to maximize throughput, adding > client threads that send Solr documents is a better approach. > > All that said, I usually just pick 1,000 and don’t worry about it. > > Best, > Erick > >> On Jul 7, 2020, at 8:59 AM, Sidharth Negi <sidharth.negi...@gmail.com> wrote: >> >> Hi, >> >> Could someone help me with the best way to go about determining the maximum >> number of docs I can send in a single update call to Solr in a master / >> slave architecture. >> >> Thanks! >
