External-Mirrors/awx
2
0
Fork 0
mirror of https://github.com/ansible/awx.git synced 2026-01-12 02:19:58 -03:30
Code Issues Packages Projects Releases Wiki Activity

Handle receptorctl advertisements for hop nodes

Browse Source 
counting it towards their heartbeat.  Also, leave off the link to the
health check endpoint from hop node Instances.
This commit is contained in:
Jeff Bradberry 2022-01-24 15:39:54 -05:00
parent 56f8f8d3f4
commit 334c33ca07
3 changed files with 12 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
awx/api/serializers.py
View File

@ -4833,7 +4833,8 @@ class InstanceSerializer(BaseSerializer):
res['jobs'] = self.reverse('api:instance_unified_jobs_list', kwargs={'pk': obj.pk})
res['instance_groups'] = self.reverse('api:instance_instance_groups_list', kwargs={'pk': obj.pk})
if self.context['request'].user.is_superuser or self.context['request'].user.is_system_auditor:
res['health_check'] = self.reverse('api:instance_health_check', kwargs={'pk': obj.pk})
if obj.node_type != 'hop':
res['health_check'] = self.reverse('api:instance_health_check', kwargs={'pk': obj.pk})
return res
def get_consumed_capacity(self, obj):

2
awx/main/models/ha.py
View File

@ -195,7 +195,7 @@ class Instance(HasPolicyEditsMixin, BaseModel):
if ref_time is None:
ref_time = now()
grace_period = settings.CLUSTER_NODE_HEARTBEAT_PERIOD * 2
if self.node_type == 'execution':
if self.node_type in ('execution', 'hop'):
grace_period += settings.RECEPTOR_SERVICE_ADVERTISEMENT_PERIOD
return self.last_seen < ref_time - timedelta(seconds=grace_period)

16
awx/main/tasks/system.py
View File

@ -436,14 +436,11 @@ def inspect_execution_nodes(instance_list):
workers = mesh_status['Advertisements']
for ad in workers:
hostname = ad['NodeID']
if not any(cmd['WorkType'] == 'ansible-runner' for cmd in ad['WorkCommands'] or []):
continue
changed = False
if hostname in node_lookup:
instance = node_lookup[hostname]
else:
logger.warn(f"Unrecognized node on mesh advertising ansible-runner work type: {hostname}")
logger.warn(f"Unrecognized node advertising on mesh: {hostname}")
continue
was_lost = instance.is_lost(ref_time=nowtime)
@ -454,6 +451,10 @@ def inspect_execution_nodes(instance_list):
instance.last_seen = last_seen
instance.save(update_fields=['last_seen'])
# Make sure that hop nodes don't fall through and have the execution_node_health_check task applied
if not any(cmd['WorkType'] == 'ansible-runner' for cmd in ad['WorkCommands'] or []):
continue
if changed:
execution_node_health_check.apply_async([hostname])
elif was_lost:
@ -482,7 +483,6 @@ def cluster_node_heartbeat():
for inst in instance_list:
if inst.hostname == settings.CLUSTER_HOST_ID:
this_inst = inst
instance_list.remove(inst)
break
else:
(changed, this_inst) = Instance.objects.get_or_register()
@ -506,7 +506,9 @@ def cluster_node_heartbeat():
raise RuntimeError("Cluster Host Not Found: {}".format(settings.CLUSTER_HOST_ID))
# IFF any node has a greater version than we do, then we'll shutdown services
for other_inst in instance_list:
if other_inst.version == "" or other_inst.version.startswith('ansible-runner') or other_inst.node_type == 'execution':
if other_inst.node_type in ('execution', 'hop'):
continue
if other_inst.version == "" or other_inst.version.startswith('ansible-runner'):
continue
if Version(other_inst.version.split('-', 1)[0]) > Version(awx_application_version.split('-', 1)[0]) and not settings.DEBUG:
logger.error(
@ -530,7 +532,7 @@ def cluster_node_heartbeat():
# * It was set to 0 by another tower node running this method
# * It was set to 0 by this node, but auto deprovisioning is off
#
# If auto deprovisining is on, don't bother setting the capacity to 0
# If auto deprovisioning is on, don't bother setting the capacity to 0
# since we will delete the node anyway.
if other_inst.capacity != 0 and not settings.AWX_AUTO_DEPROVISION_INSTANCES:
other_inst.mark_offline(errors=_('Another cluster node has determined this instance to be unresponsive'))