* Improve efficiency of preemption logic in clusters with many pending pods. ([#72895](https://github.com/kubernetes/kubernetes/pull/72895), [@bsalamat](https://github.com/bsalamat))



# v1.12.5

[Documentation](https://docs.k8s.io)

## Downloads for v1.12.5

     * purposes, the "count" field, tracking the number of elements, serves as that volatile
     * variable ensuring visibility. This is convenient because this field needs to be read in many
     * read operations anyway:
     *
     * - All (unsynchronized) read operations must first read the "count" field, and should not
     * look at table entries if it is 0.
     *

            "issue at https://github.com/tensorflow/tensorflow/issues.");
        return;
      }
      // We have a convoluted scheme here: Using the C++ graph construction API
      // to add potentially many nodes to the graph without running the checks
      // (such as uniqueness of the names of nodes) we run with other functions
      // that add a node to the graph (like TF_FinishOperation).

  |   be done with a regex, but we need inter-field validation and rules which could be
  |   dealt with by something like drools.
  |
  | o i18n: would be good to be able to have names/descriptions/specifications
  |   in as many languages as possible. (see MNG-3626)
  |
  | o annotation mechanism so that changes to the model can be accurately tracked.
  |

- Reduce API server memory when many CRDs are loaded by sharing a single etcd3 client logger across all clients ([#111649](https://github.com/kubernetes/kubernetes/pull/111649), [@negz](https://github.com/negz)) [SIG API Machinery]

*   Various performance improvements, especially for RNNs and convolutions
*   Many bugfixes
*   Nightly builds, tutorial tests, many test improvements
*   New examples: transfer learning and deepdream ipython notebook
*   Added tutorials, many documentation fixes.

## Thanks to our Contributors

This release contains contributions from many people at Google, as well as:

- Reduce API server memory when many CRDs are loaded by sharing a single etcd3 client logger across all clients ([#111650](https://github.com/kubernetes/kubernetes/pull/111650), [@negz](https://github.com/negz)) [SIG API Machinery]

### CronJob
* Add configurable limits to CronJob resource to specify how many successful and failed jobs are preserved. ([#40932](https://github.com/kubernetes/kubernetes/pull/40932), [@peay](https://github.com/peay))

### DaemonSet

  A secondary benefit of defending all users' freedom is that
improvements made in alternate versions of the program, if they
receive widespread use, become available for other developers to
incorporate.  Many developers of free software are heartened and
encouraged by the resulting cooperation.  However, in the case of
software used on network servers, this result may fail to come about.

- Reduce API server memory when many CRDs are loaded by sharing a single etcd3 client logger across all clients ([#111648](https://github.com/kubernetes/kubernetes/pull/111648), [@negz](https://github.com/negz)) [SIG API Machinery]