diff --git a/nova/compute/manager.py b/nova/compute/manager.py
index c8a4c22..d40ca77 100644
--- a/nova/compute/manager.py
+++ b/nova/compute/manager.py
@@ -1150,7 +1150,14 @@ class ComputeManager(manager.Manager):
finally:
if CONF.defer_iptables_apply: self.driver.filter_defer_apply_off()
- self._update_scheduler_instance_info(context, instances)
+ if instances:
+ # We only send the instance info to the scheduler on startup
+ # if there is anything to send, otherwise this host might
+ # not be mapped yet in a cell and the scheduler may have
+ # issues dealing with the information. Later changes to
+ # instances on this host will update the scheduler, or the
+ # _sync_scheduler_instance_info periodic task will.
+ self._update_scheduler_instance_info(context, instances)
Perhaps this is causing there to be a delay somewhere and nova is expecting an eventual consistency of some sort - but the shade test here is happening before whatever has woken up and registered something with the cell?
The only maybe relevant thing I can see is:
diff --git a/nova/ compute/ manager. py b/nova/ compute/ manager. py compute/ manager. py compute/ manager. py manager. Manager) : iptables_ apply:
self. driver. filter_ defer_apply_ off() scheduler_ instance_ info(context, instances) _instance_ info periodic task will. scheduler_ instance_ info(context, instances)
index c8a4c22..d40ca77 100644
--- a/nova/
+++ b/nova/
@@ -1150,7 +1150,14 @@ class ComputeManager(
finally:
if CONF.defer_
- self._update_
+ if instances:
+ # We only send the instance info to the scheduler on startup
+ # if there is anything to send, otherwise this host might
+ # not be mapped yet in a cell and the scheduler may have
+ # issues dealing with the information. Later changes to
+ # instances on this host will update the scheduler, or the
+ # _sync_scheduler
+ self._update_
Perhaps this is causing there to be a delay somewhere and nova is expecting an eventual consistency of some sort - but the shade test here is happening before whatever has woken up and registered something with the cell?