Add executor name to the build metric

buildman/manager/ephemeral.py

@@ -182,6 +182,7 @@ class EphemeralBuilderManager(BaseManager):
         self._build_uuid_to_info.pop(build_job.build_uuid, None)
         raise Return()
 
+      executor_name = build_info.executor_name
       execution_id = build_info.execution_id
 
       # If we have not yet received a heartbeat, then the node failed to boot in some way. We mark
@@ -194,7 +195,7 @@ class EphemeralBuilderManager(BaseManager):
         got_lock = yield From(self._take_etcd_atomic_lock('job-expired', build_job.build_uuid))
         if got_lock:
           logger.warning('Marking job %s as incomplete', build_job.build_uuid)
-          self.job_complete_callback(build_job, BuildJobResult.INCOMPLETE)
+          self.job_complete_callback(build_job, BuildJobResult.INCOMPLETE, executor_name)
 
       # Finally, we terminate the build execution for the job. We don't do this under a lock as
       # terminating a node is an atomic operation; better to make sure it is terminated than not.
@@ -540,7 +541,10 @@ class EphemeralBuilderManager(BaseManager):
                  build_job.build_uuid, job_status)
 
     # Mark the job as completed.
-    self.job_complete_callback(build_job, job_status)
+    build_info = self._build_uuid_to_info.get(build_job.build_uuid, None)
+    executor_name = build_info.executor_name if build_info else None
+
+    self.job_complete_callback(build_job, job_status, executor_name)
 
     # Kill the ephmeral builder.
     yield From(self.kill_builder_executor(build_job.build_uuid))

buildman/server.py

@@ -141,7 +141,7 @@ class BuilderServer(object):
     self._queue.extend_processing(build_job.job_item, seconds_from_now=JOB_TIMEOUT_SECONDS,
                                   minimum_extension=MINIMUM_JOB_EXTENSION)
 
-  def _job_complete(self, build_job, job_status):
+  def _job_complete(self, build_job, job_status, executor_name=None):
     if job_status == BuildJobResult.INCOMPLETE:
       self._queue.incomplete(build_job.job_item, restore_retry=False, retry_after=30)
     else:
@@ -152,7 +152,7 @@ class BuilderServer(object):
     if self._current_status == BuildServerStatus.SHUTDOWN and not self._job_count:
       self._shutdown_event.set()
 
-    _report_completion_status(build_job, job_status)
+    _report_completion_status(build_job, job_status, executor_name)
 
   @trollius.coroutine
   def _work_checker(self):
@@ -229,10 +229,10 @@ class BuilderServer(object):
     # Initialize the work queue checker.
     yield From(self._work_checker())
 
-def _report_completion_status(build_job, status):
+def _report_completion_status(build_job, status, executor_name):
   metric_queue.build_counter.Inc(labelvalues=[status])
   metric_queue.repository_build_completed.Inc(labelvalues=[build_job.namespace, build_job.repo_name,
-                                                           status])
+                                                           status, executor_name or 'executor'])
   if status == BuildJobResult.COMPLETE:
     status_name = 'CompleteBuilds'
   elif status == BuildJobResult.ERROR:

util/metrics/metricqueue.py

@@ -55,7 +55,7 @@ class MetricQueue(object):
     self.repository_build_completed = prom.create_counter('repository_build_completed',
                                                           'Repository Build Complete Count',
                                                           labelnames=['namespace', 'repo_name',
-                                                                      'status'])
+                                                                      'status', 'executor'])
 
     self.repository_count = prom.create_gauge('repository_count', 'Number of repositories')
     self.user_count = prom.create_gauge('user_count', 'Number of users')