Wait! Can Your Pod Survive a Restart?
Talkat KubeCon + CloudNativeCon Europe 2025 · on Friday, 2025-04-04· at 14:30· Level 1 | Hall Entrance N10 | Room F·
summary
Shows two live demos: an HTTP server and an Operator with leader election and proves how to get zero- or low-downtime restarts. Explains using Dockerfile STOP_SIGNAL or a preStop hook to send the right signal, differences between readiness, liveness and startup probe, and that sidecars shut down sequentially in reverse start order within a shared grace period. Quantifies leader-election: defaults 15s lease + 2s retry (worst ~17s) and how releaseOnCancel drops takeover to ~3s. Covers PDB pitfalls, need for ≥2 replicas for safe eviction, and Kubernetes 1.27 unhealthyPodEvictionPolicy.
AI-written summary · about
chapters (16)
- 0:00Pod restart challenges
- 1:07Why pod restarts matter
- 2:53Pod termination flow
- 4:40Handling stop signals
- 6:52Termination in multi-container pods
- 8:17Termination grace period limits
- 9:19HTTP traffic during restarts
- 11:06Readiness and startup probes
- 12:38Demo: traffic without downtime
- 15:08Traffic routing during restarts
- 17:16Single replica limitations
- 18:04Graceful controller restarts
- 20:14Leader election transitions
- 21:15Minimizing disruption with PDBs
- 23:21Unhealthy pod eviction
- 24:23Pod restart resilience
related talks
08m⚡ Fast Restarts, Not Just Fast Starts: Accelerating Pod Recovery
34mWhere's My Pod, Already? A Tale of Cold-Start Struggles on Kubernetes
33mPID 1, SIG Handling, Hooks & Probes: Managing Container Lifecycle Correctly
33mThe Gotchas of Zero-Downtime Traffic /w Kubernetes
37mSelf-Healing Applications with Docker and Kubernetes
27mHighly Available Services During Maintenance Events
25mCrossing the Streams! Rollout Strategies to Keep Your Users Happy!
38mHow Can Pod Start-up Be Accelerated on Nodes in Large Clusters?