Shyam Sreevalsan

cc: @ktsaou @cakrit @netdata/product

This is applicable for many metrics not just buffer cache hit ratio. Basically anything we're averaging across DBs and include inactive DBs will deflate the average. Also averaging percentage values...

cc: @papazach @cakrit @sashwathn and @juacker

Discussion happened w/ @ktsaou @ralphm @amalkov @sashwathn @hugovalente-pm @shyamvalsan and some of the points discussed: - There needs to be a general rule on how to handle metrics where aggregation...

@ktsaou @ralphm @ilyam8 and others please add your thoughts so we can take this discussion forward asynchronously.

@ilyam8 there were discussions on changing the value collected to a NULL or something indicating Invalid/Not applicable - this is why it is in the agent/collectors area now. There could...

@ilyam8 - Yes, this problem is not specific to Postgres DB cache hit ratio, it affects a lot of other metrics as well. And we should fix those too. The...

@papazach I have to disagree here. In this example, we have 4 dbs with only one of them active (with a cache hit ratio of 99%) but here's what Netdata...

@papazach for further reference, here's a Grafana dashboard using Netdata metrics which shows an accurate composite cache hit ratio (this is from our production k8s room btw). Formula is: `composite...

yeah that's probably the right way to solve this and other similar issues. We should be able to create derived metrics on the cloud, without requiring any collector level changes....