After setting strategy.type: Recreate in your Deployment, you observe that new pods start before old pods terminate. What could cause this behavior?
medium📝 Troubleshoot Q6 of 15
Kubernetes - ReplicaSets and Deployments
After setting strategy.type: Recreate in your Deployment, you observe that new pods start before old pods terminate. What could cause this behavior?
AThe Deployment has <code>maxUnavailable</code> set to a non-zero value
BThe Deployment controller is overridden by a custom controller ignoring the strategy
CThe pods have no readiness probes configured
DThe Deployment is using RollingUpdate strategy instead
Step-by-Step Solution
Solution:
Step 1: Verify Strategy Enforcement
Recreate strategy should delete old pods before creating new ones.
Step 2: Identify External Interference
If a custom controller or operator manages the Deployment, it might ignore the strategy setting.
Step 3: Exclude Other Causes
maxUnavailable is not used with Recreate; readiness probes affect pod readiness but not deletion order; using RollingUpdate would not match the given setting.
Final Answer:
The Deployment controller is overridden by a custom controller ignoring the strategy -> Option B
Quick Check:
Custom controllers can override update behavior [OK]
Quick Trick:Custom controllers may ignore Recreate strategy [OK]
Common Mistakes:
Assuming maxUnavailable applies to Recreate
Blaming readiness probes for pod deletion order
Not checking for custom controllers
Master "ReplicaSets and Deployments" in Kubernetes
9 interactive learning modes - each teaches the same concept differently
