Container Live Migration
summary
CRIU provides process-level checkpoint/restore that enables live container migration and is integrated into Podman, runc/crun, lxc/lxd, Borg, Docker and experimental cri-o/kubernetes support. Demos show kubernetes drain writing checkpoints to disk and restoring after reboot, podman creating checkpoint tar archives that include filesystem changes for easy transfer, and fast-start copies that start ~50% faster. Technical details cover injected parasite code to dump memory, kernel clone3 to reassign PIDs, and fixes for SElinux confinement to make restores work.
AI-written summary · about
chapters (8)
- 0:00Introduction and agenda
- 1:54Defining container migration
- 3:46Reboot and state demo
- 12:54Quick startup and copying
- 18:06Inside the CRIU tool
- 25:13Restore process and mechanics
- 30:02Ecosystem and migration demo
- 36:02Future outlook and summary
related talks
26mKubernetes and Checkpoint Restore
13mLightning Talk: Containers Live Migration: What’s There, What’s Missing, What’s Next?
39mKubernetes and Checkpoint/Restore
43mContainer Live Migration in Kubernetes Production Environment
37mCRI-O: Under the Hood
34mLetting Your Containers Dream of Electric Sheep
20mCRIU and SELinux
27mState of Checkpoint/Restore in Kubernetes – Now With GPUs
