1

I'm trying to implement the new Nomad Autoscaler tool for our clusters so they can scale the EC2 instances as need and we already have Prometheus configured on all nodes of the cluster so we get metrics from Nomad. However, the metrics are very unique to an EC2 instance or job allocation, below is an example.

nomad_client_allocated_memory_18d612fa_c7bb_aa80_e0ae_ac353814fce8_dev_2_service_data_ready_eligible

As you can see the metric name has the unique ID given to the node plus the cluster name and type.

Is there a way to reshape these metrics to something more usable and work well with multiple environments?

Something like this would be an ideal metric for me.

nomad_client_allocated_memory_data_ready_eligible

I had spent a day trying to find any documentation on this but so far I've only seen how I can add tags but not reshaping the metric name itself.

If anyone can point me to good documentation on this would also be very useful as I'm not very familiar with Prometheus.

Let me know if you need more information and any help is much appreciated.

Thanks in advance

Prav
  • 2,785
  • 1
  • 21
  • 30

1 Answers1

1

Set the "disable_hostname" option in the telemetry stanza of the Nomad agent's configuration to true. By default, it is false.

https://www.nomadproject.io/docs/configuration/telemetry#disable_hostname

paladin-devops
  • 121
  • 4