I'm trying to collect ES Node statistics and I'm wondering how best to
handle the fact that neither Node UUIDs or names are static/fixed/
constant. They change after the process is restarted, so it's hard
for a tool to keep track of a specific Node.
And here is why I'd like to be able to be identify a specific Node.
I'd like to collect stats for the whole Cluster (all Nodes and all
Shards) and I'd like to be able to filter any Node-specific metric by
Node. That way I could, for instance, see metric X for 2 or more
Nodes over time on the same graph. If Node identifiers change over
time, then I can't "permanently identify" any Node, which makes it
impossible to compare 2 Nodes over time.
What is the best approach one should take with this?
Should I think of ES node as purely (host+port) combination?