𝐊𝐮𝐛𝐞𝐫𝐧𝐞𝐭𝐞𝐬 𝟏.𝟑𝟒 𝐢𝐬 𝐡𝐞𝐫𝐞!

🚀 𝐊𝐮𝐛𝐞𝐫𝐧𝐞𝐭𝐞𝐬 𝟏.𝟑𝟒 𝐢𝐬 𝐡𝐞𝐫𝐞! 😀

Released last week, with a new logo and no less than 58 enhancements (stable, beta, and alpha).

👉 One of the most anticipated features: per-container restart policies.

🔹 The problem before: a Pod-level restart policy applied to all containers (Always, OnFailure, Never). This was limiting—e.g., you might want your main container to always restart, but your init container to run only once.

🔹 The solution now: with the new ContainerRestartRules, you can set a restartPolicy per container and even use restartPolicyRules to restart containers conditionally based on exit codes.

💡 𝐔𝐬𝐞 𝐜𝐚𝐬𝐞𝐬:

ML training jobs → Restart containers “in place” on retriable errors without rescheduling the entire Pod. Init containers → Run only once, while the main container can still always restart. Multi-container Pods → Each container can follow its own restart logic.

⚙️ To use it, enable the ContainerRestartRules feature gate in Kubernetes 1.34+, then set restartPolicy and restartPolicyRules in your Pod spec.

A big step forward for flexibility and fine-grained lifecycle control in Kubernetes!

Learn More Explore the following links.👇🏻

• Read the documentation for container restart policy.
• Read the KEP for the Container Restart Rules
• kubernetes v1.34