Is there a way to have the maxSurge behaviour on a statefulset like on deployments?
When the pods of my application (which is using statefulsets because it requires to have some disk cache mechanism to speed up the boostrap) is updating during a rolling update, pods are stopped one by one. The web traffic associated to the updating pod is migrated to another one. This new pod takes then too much load.
I would prefer to use maxSurge strategy during rolling update to avoid this situation. Do you know a workaround to achieve it or am I doing something wrong?