[JENKINS-23792] Do not terminate stopped EC2 nodes that were just booted

src/main/java/hudson/plugins/ec2/EC2Computer.java

@@ -106,7 +106,7 @@ public String getDecodedConsoleOutput() throws AmazonClientException {
         GetConsoleOutputRequest request = new GetConsoleOutputRequest(getInstanceId());
         return ec2.getConsoleOutput(request).getDecodedOutput();
     }
-    
+
     /**
      * Obtains the instance state description in EC2.
      *
@@ -154,6 +154,15 @@ public String getUptimeString() throws AmazonClientException, InterruptedExcepti
         return Util.getTimeSpanString(getUptime());
     }
 
+    /**
+     * Return the time this instance was launched in ms since the epoch.
+     *
+     * @return Time this instance was launched, in ms since the epoch.
+     */
+    public long launchedAtMs() throws InterruptedException {
+        return this.describeInstance().getLaunchTime().getTime();
+    }
+
     /**
      * When the agent is deleted, terminate the instance.
      */

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

@@ -140,11 +140,13 @@ private long internalCheck(EC2Computer computer) {
 
         if (computer.isIdle() && !DISABLED) {
             final long uptime;
+            final long launchedAtMs;
             InstanceState state;
 
             try {
                 state = computer.getState(); //Get State before Uptime because getState will refresh the cached EC2 info
                 uptime = computer.getUptime();
+                launchedAtMs = computer.launchedAtMs();
             } catch (AmazonClientException | InterruptedException e) {
                 // We'll just retry next time we test for idleness.
                 LOGGER.fine("Exception while checking host uptime for " + computer.getName()
@@ -190,7 +192,7 @@ private long internalCheck(EC2Computer computer) {
                 }
             }
 
-            final long idleMilliseconds = this.clock.millis() - Math.max(computer.getIdleStartMilliseconds(), uptime);
+            final long idleMilliseconds = this.clock.millis() - Math.max(computer.getIdleStartMilliseconds(), launchedAtMs);
 
 
             if (idleTerminationMinutes > 0) {

src/test/java/hudson/plugins/ec2/EC2RetentionStrategyTest.java

@@ -11,12 +11,6 @@
 import hudson.plugins.ec2.util.SSHCredentialHelper;
 import hudson.slaves.NodeProperty;
 import hudson.slaves.OfflineCause;
-import jenkins.util.NonLocalizable;
-import org.junit.Rule;
-import org.junit.Test;
-import org.jvnet.hudson.test.JenkinsRule;
-import org.jvnet.hudson.test.LoggerRule;
-
 import java.time.Clock;
 import java.time.Duration;
 import java.time.Instant;
@@ -26,18 +20,23 @@
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Collections;
+import java.util.Date;
 import java.util.List;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.logging.Level;
 import java.util.stream.Collectors;
-
+import jenkins.util.NonLocalizable;
 import static org.hamcrest.MatcherAssert.assertThat;
 import static org.hamcrest.Matchers.containsString;
 import static org.hamcrest.Matchers.equalTo;
 import static org.hamcrest.Matchers.hasItem;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertTrue;
+import org.junit.Rule;
+import org.junit.Test;
+import org.jvnet.hudson.test.JenkinsRule;
+import org.jvnet.hudson.test.LoggerRule;
 
 public class EC2RetentionStrategyTest {
 
@@ -103,6 +102,7 @@ void idleTimeout() {
             }
         };
         EC2Computer computer = new EC2Computer(slave) {
+            private final long launchedAtMs = new Date().getTime();
 
             @Override
             public EC2AbstractSlave getNode() {
@@ -114,6 +114,11 @@ public long getUptime() throws AmazonClientException, InterruptedException {
                 return ((minutes * 60L) + seconds) * 1000L;
             }
 
+            @Override
+            public long launchedAtMs() throws InterruptedException {
+                return this.launchedAtMs;
+            }
+
             @Override
             public boolean isOffline() {
                 return isOffline == null ? super.isOffline() : isOffline;
@@ -238,6 +243,28 @@ public void testTerminateOfflineComputerIfNotConnecting() throws Exception {
         assertThat(logging.getMessages(), hasItem(containsString("offline but not connecting, will check if it should be terminated because of the idle time configured")));
     }
 
+
+    /**
+     * Do not terminate na instance if a computer just launched, and the
+     * retention strategy timeout has not expired yet. The idle timeout
+     * is correctly accounted for nodes that have been stopped and started
+     * again (i.e termination policy is stop, rather than terminate).
+     */
+    @Test
+    public void testDoNotTerminateInstancesJustBooted() throws Exception {
+        logging.record(hudson.plugins.ec2.EC2RetentionStrategy.class, Level.FINE);
+        logging.capture(5);
+        final Instant inOneMinute = Instant.now().plus(Duration.ofSeconds(60));
+        // We terminate this one
+        new EC2RetentionStrategy(
+            "5",
+            Clock.fixed(inOneMinute.plusSeconds(1), zoneId),
+            inOneMinute.toEpochMilli()
+        ).check(computerWithIdleTime(0, 0, null, false));
+        assertThat("The computer is terminated, but should not be", idleTimeoutCalled.get(), equalTo(false));
+        assertThat(logging.getMessages(), hasItem(containsString("offline but not connecting, will check if it should be terminated because of the idle time configured")));
+    }
+
     @Test
     public void testInternalCheckRespectsWait() throws Exception {
         List<Boolean> expected = new ArrayList<>();