- Kubelet might fail to restart if an existing flexvolume mounted pvc contains a large number of directories, or is full. [#75019](https://github.com/kubernetes/kubernetes/pull/75019)

## Urgent Upgrade Notes

 *  **OkHttp no longer recovers from TLS handshake failures by attempting a TLSv1 connection.**
    The fallback was necessary for servers that implemented version negotiation incorrectly. Now
    that 99.99% of servers do it right this fallback is obsolete.
 *  Fix: Do not honor cookies set on a public domain. Previously a malicious site could inject
    cookies on top-level domains like `co.uk` because our cookie parser didn't honor the [public

}

// xlMetaV2Version describes the journal entry, Type defines
// the current journal entry type other types might be nil based
// on what Type field carries, it is imperative for the caller
// to verify which journal type first before accessing rest of the fields.
type xlMetaV2Version struct {

	for i := 0; i < len(bufsizes); i++ {
		for j := 0; j < len(bufsizes); j++ {
			nwrite := bufsizes[i]
			bs := bufsizes[j]

			// Write nwrite bytes using buffer size bs.
			// Check that the right amount makes it out
			// and that the data is correct.

			w.Reset()
			buf := NewWriterSize(w, bs)
			context := fmt.Sprintf("nwrite=%d bufsize=%d", nwrite, bs)
			n, e1 := buf.Write(data[0:nwrite])

    Potential impacts:

- Fixed a regression since 1.24 in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins).
  The incorrect loop logic might lead to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which is unexpected. ([#122371](https://github.com/kubernetes/kubernetes/pull/122371), [@caohe](https://github.com/caohe)) [SIG Scheduling]

- Failed pods associated with a job with `parallelism = 1` are recreated by the job controller honoring exponential backoff delay again. However, for jobs with `parallelism > 1`, pods might be created without exponential backoff delay. ([#115020](https://github.com/kubernetes/kubernetes/pull/115020), [@nikhita](https://github.com/nikhita)) [SIG Apps]

  Previously, for IPv6, if the podSubnet had a mask lower than /112, kubeadm calculated a node-mask to be multiple of eight and splitting the available bits to maximise the number used for nodes. ([#95723](https://github.com/kubernetes/kubernetes/pull/95723), [@aojea](https://github.com/aojea)) [SIG Cluster Lifecycle]

      scope, OutputsFromTFOutputs(parent_inputs, num_loop_vars), cond_fn,
      body_fn, params->name, &loop_outputs);

  // Update name_map with newly-created ops.
  // TODO(skyewm): right now BuildWhileLoop() may alter the graph if it returns
  // a bad status. Once we fix this, we may want to return early instead of
  // executing the following code.

                              &Deallocator, &deallocator_called);

  TF_Tensor* o = TF_TensorMaybeMove(t);
  ASSERT_TRUE(o == nullptr);  // It is unsafe to move memory TF might not own.
  TF_DeleteTensor(t);
  EXPECT_TRUE(deallocator_called);
}

TEST(CAPI, LibraryLoadFunctions) {
  // TODO(b/73318067): Fix linking for the GPU test generated by the