Today our cluster went down and gcloud console output looked like this:

]$ gcloud container clusters list
NAME                 ZONE            MASTER_VERSION  MASTER_IP   MACHINE_TYPE   NODE_VERSION  NUM_NODES  STATUS
xxxxxxxxxxxxxxxxxxx  europe-west1-d  1.6.11-gke.0    xx.xx.xx.x  n1-standard-1  1.6.4 *       2          ERROR

As one can see, the kubernetes master version differs from the version of the node pool. This introduced a conflict within the cluster. The conflict could be solved after upgrading the node pool to the proper version.

What I am trying to understand now is: Why was the master on a different version in the first place. Is there an automatic update policy, that we should have known about? But if the answer is "yes": why was the node pool not upgraded accordingly?