Endpoints have an optimized code path, as they are by far the most frequently updated resource - in a steady cluster, this will often be the *only* change, caused by scale up/down.

As a result, they do not go through `PushContext`, and changes do not trigger a `PushContext` recomputation. Instead, the current state is incrementally computed based on events from `ServiceDiscovery`.

#### Conclusion

Overall, the high level config ingestion flow:

```mermaid
graph TD

        // TODO: we should be able to inject a Map<String, LifecycleMapping> directly,
        // however, SISU visibility filtering can only happen when an explicit
        // lookup is performed. The whole problem here is caused by "project extensions"
        // which are bound to a project's classloader, without any clear definition
        // of a "project scope"
        LifecycleMapping lifecycleMapping =

              return;
            }
          }
          // Remove the interrupt bit before each task. The interrupt is for the "current task" when
          // it is sent, so subsequent tasks in the queue should not be caused to be interrupted
          // by a previous one in the queue being interrupted.
          interruptedDuringTask |= Thread.interrupted();
          try {
            task.run();

              return;
            }
          }
          // Remove the interrupt bit before each task. The interrupt is for the "current task" when
          // it is sent, so subsequent tasks in the queue should not be caused to be interrupted
          // by a previous one in the queue being interrupted.
          interruptedDuringTask |= Thread.interrupted();
          try {
            task.run();

SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

   it throws an `InterruptedIOException`.

## Version 1.5.2

_2014-03-17_

 * Fix bug where deleting a file that was absent from the `HttpResponseCache`
   caused an IOException.
 * Fix bug in HTTP/2 where our HPACK decoder wasn't emitting entries in
   certain eviction scenarios, leading to dropped response headers.

## Version 1.5.1

_2014-03-11_

   *
   * <pre>
   * com....PotentialDeadlockException: Potential Deadlock from LockC -&gt; ReadWriteA
   *   at ...
   *   at ...
   * Caused by: com...ExampleStackTrace: LockB -&gt; LockC
   *   at ...
   *   at ...
   * Caused by: com...ExampleStackTrace: ReadWriteA -&gt; LockB
   *   at ...
   *   at ...
   * </pre>
   *

    it.next();
    it.next();
    mmHeap.remove(4);
    assertThrows(ConcurrentModificationException.class, () -> it.next());
  }

  /** Tests a failure caused by fix to childless uncle issue. */
  public void testIteratorRegressionChildlessUncle() {
    final ArrayList<Integer> initial = Lists.newArrayList(1, 15, 13, 8, 9, 10, 11, 14);

have different security properties and provide different levels of confidence
when dealing with untrusted data. Based on the security history of these
libraries we consider that it is safe to work with untrusted inputs for PNG,
BMP, GIF, WAV, RAW, RAW\_PADDED, CSV and PROTO formats. All other input formats,
including tensorflow-io should be sandboxed if used to process untrusted data.

For example, if an attacker were to upload a malicious video file, they could

		// the CNI binary installed.
		if err = installer.Run(ctx); err != nil {
			if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) {
				log.Infof("installer complete: %v", err)
				// Error was caused by interrupt/termination signal
				err = nil
			} else {
				log.Errorf("installer failed: %v", err)
			}
		}

		if cleanErr := installer.Cleanup(); cleanErr != nil {
			if err != nil {