-
Story
-
Resolution: Done
-
Major
-
None
-
None
-
None
Introduction
Common use cases around collecting debugging data via eBPF can be either done from a privileged debug container, or a dedicated tool on the node. In the latter case we could collect the data from an invocation of must-gather.
Origin of this story were the discussions about debugging use cases with folks from the eBPF mail thread:
https://docs.google.com/document/d/1rXVr_O1y8L4SuJDVQZhR41kYGEOj1H1XVc7-IRTv0Rg/edit#heading=h.m751e7zgu5w1
User Story
As a Software Engineer, I'd like to know eBPF based debugging use cases, so that I can decide where a possible implementation should live.
Additional resources
- OpenShift eBPF Use Cases
- Troubleshooting script that were used in practice to find k8s container latency issue:
https://github.com/xdp-project/xdp-project/blob/master/areas/latency/softirq_net_latency.bt
Related to GitHub’s k8s issues:
https://github.blog/2019-11-21-debugging-network-stalls-on-kubernetes/
- relates to
-
RFE-2112 Need RHCOS Metrics Exposed for OCP Cluster Monitoring
- Rejected