In 5.6.5, the monitoring overview shows me point-in-time CPU across the cluster, and the node views show me historical CPU for a single node. Can I get a cluster-wide historical view?
Background ... we have a single daily index that is almost write only. So the servers that host shards of that index are much busier than servers that only host shards from prior days. Having a CPU visualization with 10 servers busy and the rest idle would help people understand that we cannot scale by adding servers unless we change the number of shards in the current day index.
We have plans for providing something along these lines, but it's unfortunately not a one-size-fits-all type of problem. To display such a chart would require a date histogram with an inner terms aggregation across all nodes and that does not scale after a reasonable number of nodes.
My current thinking is that we will provide a mechanism with something like the Waffle Map / Chart (as noted in our Elastic{ON} 2018 Keynote -- the video will be online soon if it's not already) as well as so-called Sparklines (a small chart that appears in the tabular listings per node) that could very easily be visually compared.
Hope that provides some insight into where we're planning on going!
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.