More logging when we decide to time out a slave.

Summary:
They're timing out much more often than I'd expect -- even when we're
doing back-to-back deploys they are timing out.  Not sure why.  I am
hoping this logging will shed some light.

Test Plan: Fingers crossed.

Reviewers: chris

Reviewed By: chris

Differential Revision: http://phabricator.khanacademy.org/D11464

src/main/java/hudson/plugins/ec2/EC2RetentionStrategy.java

@@ -69,7 +69,7 @@ public synchronized long check(EC2Computer c) {
             // TODO: really think about the right strategy here
             final long idleMilliseconds = System.currentTimeMillis() - c.getIdleStartMilliseconds();
             if (idleMilliseconds > TimeUnit2.MINUTES.toMillis(idleTerminationMinutes)) {
-                LOGGER.info("Idle timeout: "+c.getName());
+                LOGGER.info("Idle timeout: " + c.getName() + " after " + idleTerminationMinutes + " minutes (idle since time_t " + (int)(c.getIdleStartMilliseconds() / 1000) + ")");
                 c.getNode().idleTimeout();
             }
         }