Mix Kubernetes with Linux Kernel Tracing to Boost the Observability of Your Cluster – Tzvetomir Iliev Stoyanov, VMware
The Linux kernel has rich tracing capabilities, built-in and enabled by default in almost all of the major Linux distributions. They evolved in the past 30 years from a simple set of printk()s to a highly optimized framework, capable to trace every corner of the kernel running on a production system. Originally designed for a kernel tracing, to help kernel developers debug their code, now this framework is extended and can be useful for user space tracing too. The majority of Kubernetes clusters run on Linux and all these tracing benefits are already there, waiting to be used. Kubernetes pods are just an abstraction based on Linux namespaces, from the kernel point of view they are regular user space processes that can be traced. In this talk Tzvetomir will describe a newly originated open source project, that leverages Linux Kernel tracing framework for tracing containers running in a Kubernetes cluster.
