Ja, it’s basically a pipe on steroids, so many inputs/outputs/filters/processors etc
I used CW Exporter at last job. Was mainly to get ELB metrics, the downside is CW is delayed by 15-20 mins usually, so something to think about if you want to then alert on those metrics :(
what part of CW was delayed - the metrics ingest? or the alarming? i haven’t noticed any delays with native metrics so i’m curious because i was thinking of going down the same path for our k8s stuff…
wow i completely missed this - but CW metrics ingest was delayed, I think the issue is due to the specific metrics I was after having a high resolution - best explanation I found was from datadog: https://docs.datadoghq.com/integrations/faq/cloud-metric-delay/
Datadog, the leading service for cloud-scale monitoring.
ohh right, ok that makes sense - in the context of getting the metrics out to another service. although - that delay is still a lot. but, thanks for the context!
https://github.com/influxdata/telegraf is a pretty cool swiss army knife for this stuff
Thanks Josh, I didn’t know telegraf can have a prometheus compatible