Monitor running processes and their current state using a heat map

[joakim_berg]

Hi,

Im using a heat map visualization to show if a a set of specific processes are running and their current state "sleep" or "running".

This scales fairly well and I can see if these 4-5 processes are running on 10-20 servers showing green boxes if they are present or empty ones if they are not present.

I have some issues in that Im using Count aggregation and what I really want to do is to show if "myprocess" exist in the latest "system.process.cmdline" then display a green box, if not it can be empty or turn red.

Right now Im setting the timespan to 60 seconds and then I get various different counts for the processes( this is ok as anything except for 0 is good) but sometimes its empty even though the process is running. I would expect count to always be > 1 and then the number would vary depending on how many runs it made within that timeframe.

Do you have any ideas why?

[Joe_Fleming]

Right now Im setting the timespan to 60 seconds

You mean you're setting the time window in Kibana to 60 seconds? How often are you indexing the data? And how busy is the cluster? It can take a little time for Elasticsearch to index the data and make it available for searching. 60 seems like plenty of time, but if the cluster is under a lot of load, maybe it's not (I still think it is...). However, if you're only indexing new documents every 60 seconds, you could just be hitting the window where the data says it's indexed, but it isn't yet ready for searching.