Unable to mount volumes for pod. Timeout waiting for mount paths to be created.
1 view
Skip to first unread message
Lukas Sägesser
May 20, 2016, 5:07:08 AM5/20/16
to Containers at Google
Suddenly a Mongo DB pod that has been running for weeks tries to restart and the data disk can't be mounted anymore.
This error repeats without end:
2016-05-20 10:51:49 +0200 CEST 2016-05-20 10:53:31 +0200 CEST 2 mongo-controller-1g2xh Pod Warning FailedMount {kubelet gke-test-cluster-default-pool-a7afbbdb-i0sb} Unable to mount volumes for pod "mongo-controller-1g2xh_default(9838e85c-1e67-11e6-8ecf-42010af000c0)": Could not attach GCE PD "test-mongo-disk-test". Timeout waiting for mount paths to be created.
When I ssh into the node, "mount | grep mongo" gives no result.
But in the GCE Console in my browser, that node lists the disk as connected.
How does this happen? Why doesn't the node recover on its own?
This error repeats without end:
2016-05-20 10:51:49 +0200 CEST 2016-05-20 10:53:31 +0200 CEST 2 mongo-controller-1g2xh Pod Warning FailedMount {kubelet gke-test-cluster-default-pool-a7afbbdb-i0sb} Unable to mount volumes for pod "mongo-controller-1g2xh_default(9838e85c-1e67-11e6-8ecf-42010af000c0)": Could not attach GCE PD "test-mongo-disk-test". Timeout waiting for mount paths to be created.
When I ssh into the node, "mount | grep mongo" gives no result.
But in the GCE Console in my browser, that node lists the disk as connected.
How does this happen? Why doesn't the node recover on its own?
Lukas Sägesser
May 20, 2016, 5:42:26 AM5/20/16
to Containers at Google
Found this and posted my workaround while waiting for the fix to be released: https://github.com/kubernetes/kubernetes/issues/14642#issuecomment-220561817
Reply all
Reply to author
Forward
0 new messages