maxSurge

maxSurge is a parameter used in the rolling update strategy of Kubernetes Deployments. It controls the maximum number of additional pods that can be created above the desired replica count during a rolling update, allowing the new version to be brought up while old pods are replaced.

The value of maxSurge can be specified as an integer or as a percentage (for example, "2" or "25%"). It is expressed in the same unit as the overall replica count. During a rollout, the total number of pods can temporarily exceed the desired number of replicas by up to maxSurge, while the system tries to keep at least the desired number minus maxUnavailable pods available. In other words, the pod count during an update is constrained to a range roughly between (desired replicas − maxUnavailable) and (desired replicas + maxSurge).

Defaults and interaction with other settings: If maxSurge is not specified, Kubernetes uses a default value (commonly a percentage of the desired replicas, often 25%). This is typically considered alongside maxUnavailable, which defines how many pods may be unavailable during the update. Together, maxSurge and maxUnavailable determine both the pace of the rollout and the minimum level of availability during updates.

Practical use: Adjust maxSurge to balance upgrade speed against resource usage and risk. A higher surge allows faster rollouts but requires more cluster capacity, while a lower surge reduces peak resource demand but may slow the update. Proper tuning depends on workload characteristics and available compute resources.