SIG-Node, with Dawn Chen

Show notes

Dawn Chen, TL for SIG-Node and the Google Kubernetes Engine node team, joins Craig and Adam this week. She has worked on containers and container schedulers since 2007 - not a typo. We also bring you the news, in part from the echo chamber of Google Cloud Summit in Sydney. Do you have something cool to share? Some questions? Let us know: web: kubernetespodcast.com mail: kubernetespodcast@google.com twitter: @kubernetespod Chatter of the week Stickers! Google Cloud Summit in Hong Kong Google Cloud Next in London Gartner Symposium Orlando KubeCon Shanghai News of the week NetApp acquires StackPointCloud Cloud news: Sandbox pods on GKE Signup form Kubernetes tools for Azure Stack EKS can generate kubeconfig files! GSoC: katacontainer support in containerd, by Jian Liu linkerd 2.0 is GA Thomas Rampelberg tells you how to use it Cortex added to the CNCF sandbox Red Hat OpenShift Service Mesh, based on Istio Microservice observability with Istio at Trulia Contour 0.6 from Heptio Links from the interview Dawn Chen on GitHub The Borg paper Process containers (later 'cgroups'): The first submission of containers to the Linux kernel Early coverage of process containers Paul Menage's 2007 paper "Adding Generic Process Containers to the Linux Kernel" Dawn's first job: tracking processes. Each job had its own GID - she would use netlink connection tracking to map processes and threads to GIDs, and, using procfs, figure out CPU and memory usage. Dawn's second job: adjusting CPU usage using nice Today we just use memcg Fake NUMA - cut a machine into big chunks and assign them to groups of processes. Linux Plumbers Conference Tim Hockin's presentation at the Linux Plumbers Conference in 2011, talking about the work Dawn's team were doing lmctfy - Let Me Contain That For You In case you don't get the joke It's like runc and containerd SIG Node Node and lifecycle management Application management Container runtimes and kubelet Node problem detection Resource management GPU & TPU Security isolation gVisor and Sandbox Pods Logging and monitoring Was SIG Node the first SIG? Tied with SIG API Machinery How did we get to CRI? rktnetes was released with Kubernetes 1.3 Hyper_ containers (now Kata Containers) LXC and LXD kubevirt for running VMs instead of containers OCI CRI was released with Kubernetes 1.5 containerd and CRI-O Container RuntimeHandler, so some pods can run with one runtime and some with another

Guests

Dawn Chen

Tech Lead

Google