问题
I finally got Micrometer(and it's JVM metrics) working in Google cloud by making the instance of cloud run of type Generic Task which allows me then to focus on the whole 'named' cluster of things so I can separate out production, or other clusters since they are called prod_name and staging_name. Now, it seems micrometer is reporting metrics with tags (instead of just changing the name of the metric) and google can't seem to separate out those tagged metrics to aggregate that stat in GCP. I have 2 questions really
- Can google aggregate the metrics across instances with the metric labels such that I can use MAX(worst case instance)?
- Can I modify micrometer in some standard way to report differently so that I can aggregate things?(as a workaround to this odd google design)
Here is the picture and you will see TWO instances with postfixes 00053-zib and 00054-pur. Normally, I want to know the MAX memory used(worst instance we have), but how can I aggregate these as it seems in google, I can only aggregate over the resource labels not metric labels?
More specifically, I want the MAX of all "Tenured Gen" to see the worst case node. (I am still trying to figure out total memory usage from micrometer if it even reports that stat...I haven't found that stat yet). That's a question I am still researching.
来源：https://stackoverflow.com/questions/60728990/google-cloud-metrics-and-micrometer-jvm-reporting-is-this-a-micrometer-bug-or