Maybe I am missing something and my use case is a bad idea, but I really love the visualization tools directly built into kibana, and was hoping to use a kibana dashboard as a radiator on both developer machines, as well as on raspberry pi (2s) running directly connected to TVs in our office.
On a developer machine (recent laptop), this runs quick and snappy and really is useful.
However on the raspberry pi it's way too slow to use. I am running a raspberry pi with latest Raspbian release, with everything up to date. I have tried this on chromium as well as other browsers, but it takes 5 minutes or more just to load the kibana main page regardless.
I have created an index pattern that is restricted to just the last few days of data (my data set is not that huge to begin with), and I have a dashboard with 6 tables with the last 6 results in each, each cut to their own view of the data. If I set this to autoupdate every minute or so, I am always multiple minutes behind and often seeing chromium warnings for "this page has become unresponsive".
Is there any other way I can make this faster? Yes I know I am not posting inner details of my indices, data volume, etc, but the difference is 100000x for something that at face value is relatively simple.
On a separate local network server with solid specs. As I said it all runs great when I access the dashboard from a developer laptop. I actually have two dashboards, with one querying against an index pattern that goes back a full year's worth of data. Even with that range of data on a developer laptop it takes 1 second to load. This dashboard also has about 12 elements including 6 tables and 6 graphs
If I strip down the data to a bare minimum with an index that only goes back 10 days, and with a dashboard only shows 5 of the tables (the 6th queries against a more dense dataset), then as I said Kibana still takes more than 5 minutes to load into period on my Pi, and then still takes about 1-2 minutes to update (on a 1 minute update schedule).
I understand the Pi is a weak setup, but I wasn't expecting it to lag by this big of factor behind. In comparison we have an old static radiator that loads in a couple of seconds.
If I check the Pi's resource stats, the ram is not maxing out (about 60% used), and Chromium is distributing it's workload across all 4 cores.
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.