We’ve just seen a task failure with

containers with unready status: [fv6d6b4u4zgwuy-n10-0]|context deadline exceeded

- the task has not been retried (it would have three user retires set). Is this expected?

We are not 100% that this is the fix, but we saw the number of occurrences of this error go significantly down after increasing our kublet registry pull limits:

registryPullQPS=50

and

registryBurst=100

if you describe the pod is there any more information? any logs from propeller for that pod?