> Solr reports a memory usage of nearly 100%. This, I believe, is good as that indicates full memory utilization by MMapDirectory.
In case Solr is saying that 100% of the 2GB heap memory is used, it is not a good thing. On Fri, 9 Feb, 2024, 5:26 pm Jan Høydahl, <jan....@cominvent.com> wrote: > Hi, > > You may want to upload those screenshots to some external service as they > did not display on the list. > > Solr uses Java APIs to ask for system memory, and when running in a > container which is resource limited, it will show the max limit allocated > to that container. > > Jan > > > 9. feb. 2024 kl. 12:02 skrev Stian Brattland <st...@octetnest.no>: > > > > Hello > > > > I'm running Solr as an AWS Fargate task. The Fargate task has 8GB memory > to its availability. > > > > The Solr process has Xms=512 and Xmx=2048. The remaining memory should > go to MMapDirectory for index caching. > > > > Solr reports a memory usage of nearly 100%. This, I believe, is good as > that indicates full memory utilization by MMapDirectory. > > > > However, AWS Fargate reports that only ~22% of the task's available > memory is used. This percentage "nearly" aligns with the Xmx set for the > container, but not the full memory expected usage. > > > > I'm curious why Solr and AWS disagrees on the memory usage. It seems > that AWS only reports the direct memory usage of the container process > (being Solr), but I'm not sure. > > > > Does anyone know which system metric Solr takes the Physical Memory > from, as shown in the top bar of the below picture? > > > > Kind regards, > > Stian Brattland > > > > Solr : > > > > AWS : > > > > > >
