Kibana APM Metrics: "CPU usage" and "Memory usage" graphs do not show data when search criteria is active

Kibana version: 6.7.1

Elasticsearch version: 6.7.1

APM Server version: 6.6.1

Browser version: Chrome 74.0.3729.131 (Official Build) (64-bit), user agent: Mozilla/5.0 (iPhone; CPU iPhone OS 11_0 like Mac OS X) AppleWebKit/604.1.38 (KHTML, like Gecko) Version/11.0 Mobile/15A372 Safari/604.1

Kibana APM Metrics: "CPU usage" and "Memory usage" graphs do not show data when search criteria is active:
Here is the APM Metrics tab without active search criteria

and here is the APM Metrics tab with active search criteria

Steps to reproduce:

  1. open kibana
  2. navigate to APM
  3. select a service
  4. click on the Metrics tab
  5. notice the CPU usage and Memory usage graphs have data
  6. enter some search criteria
  7. notice the CPU usage and Memory usage graphs have data (yet the Transaction duration and Transactions per minute graphs do have data

Hi and thanks for your question. How are you adding the environment tag to the transaction and which agent are you using? Some of our agents have an environment configuration option. Setting this option should also carry through in the metrics but setting a tag on the transactions does not influence the metrics.

I use the addLabels() function on the transaction object of the elastic-apm-node agent to set the environment and the (Azure) region.

why shouldn't CPU usage and Memory usage stats have the context tags from the transaction?

Because they are scoped to only that one particular transaction and not the whole system.
I think the Node agent does not support the environment config just yet.

I would love to have labels/tags at the service level instead of the transaction level :slight_smile:

When I have multiple environments/regions emitting APM data into the same elastic server, how can I segment them when looking at their CPU and Memory usage? I am supplying the same labels/tags to every transaction in the service.

You said "...and not the whole system". Could not the same logic be used to not show CPU and Memory usage at the service level? Aren't those metrics at the machine level? What if more than one APM-monitored service is running on the same machine simultaneously? Or other non-APM'ed CPU and Memory resource users?

We actually have upcoming support for configuring global labels. However, it requires APM Server 7.2 which is not released yet. For more information see https://github.com/elastic/apm-agent-nodejs/issues/963

Thanks for the info. I am now eagerly awaiting elastic 7.2.

This topic was automatically closed 20 days after the last reply. New replies are no longer allowed.