* GCI: Update the command to get the image ([#24987](https://github.com/kubernetes/kubernetes/pull/24987), [@andyzheng0831](https://github.com/andyzheng0831))
* Port-forward: use out and error streams instead of glog ([#17030](https://github.com/kubernetes/kubernetes/pull/17030), [@csrwng](https://github.com/csrwng))

     com.squareup.okio:okio:1.14.0
     ```

 *  Fix: Handle `HTTP/1.1 100 Continue` status lines, even on requests that did
    not send the `Expect: continue` request header.
 *  Fix: Do not count web sockets toward the dispatcher's per-host connection
    limit.
 *  Fix: Avoid using invalid HTTPS sessions. This prevents OkHttp from crashing
    with the error, `Unexpected TLS version: NONE`.

  optional int64 seconds = 1;

  // Non-negative fractions of a second at nanosecond resolution. Negative
  // second values with fractions must still have non-negative nanos values
  // that count forward in time. Must be from 0 to 999,999,999
  // inclusive. This field may be limited in precision depending on context.
  optional int32 nanos = 2;
}

# This is the same as the official TensorFlow builds.
# See https://developer.nvidia.com/cuda-gpus#compute
# `compute_XY` enables PTX embedding in addition to SASS. PTX
# is forward compatible beyond the current compute capability major
# release while SASS is only forward compatible inside the current
# major release. Example: sm_80 kernels can run on sm_89 GPUs but
# not on sm_90 GPUs. compute_80 kernels though can also run on sm_90 GPUs.