👋 New to SigNoz and feeling out the cloud offering... I created https://github.com/SigNoz/charts/issues/763, which, as I worked on the problem, started to feel more like a general support request, so I thought I'd come here instead. So... 1. I'm using GKE Autopilot and Signoz's k8s infra helm chart (0.14.1) 2. I had to

--set presets.hostMetrics.enabled=false

to get the otel-agent (daemon-set) deployment to stop crash looping 3. Log collection is working, but only for the signoz pods. 4. the otel-agent deployment can't seem to successfully read some of the k8s APIs (

/pods

in particular is failing) I would love to understand how I can 1. enable host metrics 2. expand log collection 3. get k8s API collection working I know this might come off as a "please do my homework for me" style request, but I thought I would ask early in my debugging in case someone else had already seen similar issues and had a quick answer... Thanks, all.

@Nagesh Bansal please help out with gke autopilot collection agent.

I think it's worth asking: Do we need the daemonset in AutoPilot clusters? Or can we just use the single instance for cluster collection?

Hi @John Seekins - I don't work for Signoz, but do have log collection and host metrics working with a couple of GKE autopilot clusters, including non-signoz pods. For log collection - did you set up auto instrumentation in the non-signoz pods? https://signoz.io/docs/instrumentation/ How easily this works tended to change for me depending on which framework was being instrumented. Debugging steps when setting this up were. 1. Make sure the signoz otel collector is running. 2. Make sure your services logs are being generated with kubectl. 3. Verify services have correct OTLP endpoint. 4. Confirm Signoz Otel collector is receiving them, and they exist in the Clickhouse DB. Host metrics - I did deploy a daemonset to get this working. This was helpful as a reference - https://signoz.io/blog/using-signoz-to-monitor-your-kubernetes-cluster/

I was kinda hoping to follow these docs for log collection: https://signoz.io/docs/gcp-monitoring/cloud-run/logging/ If you take out the filter in the pub/sub topic, you can just consume all the logs through one topic.