forked from rook/rook
Satoru Takeuchi
06b10e248c
Support topologySpreadConstraints to spread OSD pods among failure domains as even as possible. https://kubernetes.io/docs/concepts/workloads/pods/pod-topology-spread-constraints/ Without this feature, the number of OSD pods would be different between the failure domains very much. At worst, there is a possibility that a failure domain has many OSDs and other failure domains have no OSDs. It can partly be mitigated with podAntiAffinity. However, it only works if there are at most one OSD per node. For more detail, please refer to the following document. https://github.com/rook/rook/blob/master/cluster/examples/kubernetes/ceph/cluster-on-pvc.yaml#L59 Closes: https://github.com/rook/rook/issues/4387 Signed-off-by: Hiroshi Muraoka <h.muraoka714@gmail.com> Signed-off-by: Satoru Takeuchi <satoru.takeuchi@gmail.com>