Merge pull request #13608 from AlanCoding/keepalive

Use ansible-runner change to get periodic keep-alive messages in K8S

awx/main/conf.py

@@ -282,6 +282,16 @@ register(
     placeholder={'HTTP_PROXY': 'myproxy.local:8080'},
 )
 
+register(
+    'AWX_RUNNER_KEEPALIVE_SECONDS',
+    field_class=fields.IntegerField,
+    label=_('K8S Ansible Runner Keep-Alive Message Interval'),
+    help_text=_('Only applies to jobs running in a Container Group. If not 0, send a message every so-many seconds to keep connection open.'),
+    category=_('Jobs'),
+    category_slug='jobs',
+    placeholder=240,  # intended to be under common 5 minute idle timeout
+)
+
 register(
     'GALAXY_TASK_ENV',
     field_class=fields.KeyValueField,

awx/main/tasks/callback.py

@@ -85,6 +85,8 @@ class RunnerCallback:
         # which generate job events from two 'streams':
         # ansible-inventory and the awx.main.commands.inventory_import
         # logger
+        if event_data.get('event') == 'keepalive':
+            return
 
         if event_data.get(self.event_data_key, None):
             if self.event_data_key != 'job_id':

awx/main/tasks/receptor.py

@@ -526,6 +526,10 @@ class AWXReceptorJob:
         pod_spec['spec']['containers'][0]['image'] = ee.image
         pod_spec['spec']['containers'][0]['args'] = ['ansible-runner', 'worker', '--private-data-dir=/runner']
 
+        if settings.AWX_RUNNER_KEEPALIVE_SECONDS:
+            pod_spec['spec']['containers'][0].setdefault('env', [])
+            pod_spec['spec']['containers'][0]['env'].append({'name': 'ANSIBLE_RUNNER_KEEPALIVE_SECONDS', 'value': str(settings.AWX_RUNNER_KEEPALIVE_SECONDS)})
+
         # Enforce EE Pull Policy
         pull_options = {"always": "Always", "missing": "IfNotPresent", "never": "Never"}
         if self.task and self.task.instance.execution_environment: