You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm trying to get the k8sattributes processor working with beyla but am having issues where traces from pods lose their annotations and metadata once the pod has been rotated.
What I've worked out is that once the pods are restarted beyla.ebpf stops adding k8s metadata to the resource attributes and then the k8sattribute processor can't match up the spans to pods.
I've been testing this using Kind and Alloy is running as a daemonset.
No errors are reported in the logs and Alloy has enough permissions.
Just seems that as soon as some pods are rotated that alloy forgets to add the metadata etc.
I'm hoping this is just an issue with my configuration but any help would be appreciated.
Steps to reproduce
Run a kind cluster
Run a random deployment with some annotations.
Create alloy daemonset using the below config.
Once the alloy daemonset is running, rollout restart the deployment.
matt-gp
changed the title
otelcol.processor.k8sattributes
otelcol.processor.k8sattributes - New Pods Missing Metadata Labels
Oct 31, 2024
matt-gp
changed the title
otelcol.processor.k8sattributes - New Pods Missing Metadata Labels
otelcol.processor.k8sattributes - New Pods Missing Resource Attribute Labels
Oct 31, 2024
matt-gp
changed the title
otelcol.processor.k8sattributes - New Pods Missing Resource Attribute Labels
beyla.ebpf - New Pods Missing Resource Attribute Labels
Nov 10, 2024
What's wrong?
Hi all,
I'm trying to get the k8sattributes processor working with beyla but am having issues where traces from pods lose their annotations and metadata once the pod has been rotated.
What I've worked out is that once the pods are restarted
beyla.ebpf
stops adding k8s metadata to the resource attributes and then the k8sattribute processor can't match up the spans to pods.I've been testing this using Kind and Alloy is running as a daemonset.
No errors are reported in the logs and Alloy has enough permissions.
Just seems that as soon as some pods are rotated that alloy forgets to add the metadata etc.
I'm hoping this is just an issue with my configuration but any help would be appreciated.
Steps to reproduce
System information
linux/arm64/kubernetes:v1.29.2
Software version
grafana/alloy:v1.4.3
Configuration
Logs
Before the deployment is restarted
After rotation
The text was updated successfully, but these errors were encountered: