[FIXED JENKINS-23850] Workaround AWS bug that hides stopped slaves

Use instance.group-name instead of group-name as the describeInstances filter when searching for stopped slaves to restart. Some EC2 regions are incorrectly returning empty describeInstances results when the group-name filter is used. This causes jenkins to start new slaves instead of restarting stopped instances. instance.group-name is functionally identical and seems to work consistently. https://issues.jenkins-ci.org/browse/JENKINS-23850
jenkinsci · Jul 18, 2014 · 14d537a · 14d537a · justinmaiuto · Jul 30, 2014
1 parent 6347c08
commit 14d537a
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/src/main/java/hudson/plugins/ec2/SlaveTemplate.java b/src/main/java/hudson/plugins/ec2/SlaveTemplate.java
@@ -319,7 +319,7 @@ private EC2AbstractSlave provisionOndemand(TaskListener listener) throws AmazonC
                /* No subnet: we can use standard security groups by name */
                 riRequest.setSecurityGroups(securityGroupSet);
                 if (securityGroupSet.size() > 0)
-                    diFilters.add(new Filter("group-name").withValues(securityGroupSet));
+                    diFilters.add(new Filter("instance.group-name").withValues(securityGroupSet));
             }
 
             String userDataString = Base64.encodeBase64String(userData.getBytes());