New Relic does not allow alerting on Prometheus network metrics

the following prometheus metrics are accumulative over time:

node_network_receive_fifo_total
node_network_transmit_fifo_total
node_network_transmit_packets_total
node_network_receive_packets_total
node_network_transmit_errs_total
node_network_receive_errs_total
node_network_transmit_drop_total

for example:

our main goal was to alert on those intervals were dropped (just to say an example) packages happened, via NRQL we are able to obtain those periods just fine, with different queries:

  • SELECT latest(node_network_receive_drop_total) - earliest(node_network_receive_drop_total) as ‘dropped’ from Metric facet instance,device where instrumentation.source = ‘systemx’

  • SELECT stddev(node_network_receive_drop_total) FROM Metric WHERE instrumentation.source = ‘systemx’ FACET instance,device TIMESERIES

however, when we try to create an alert out of them, the alert wizard basically does not show anything on the values (like if the query were invalid)

Promql offers great flexibility however New Relic has huge limitations when it comes to alerting based on prometheus data, any idea how I can get around this?

I think I managed to get this working

Hi @robert.sanchez

Thanks for reaching out, I hope you are well.

Congrats on your first post in the community, Whoop!

Great to hear you go this working, please feel free to share your workaround / fix. As it may be helpful to other users facing the same issue.