Exception no valid host when try launch an instance
Hello,
I setup openstack rocky version with three nodes (compute, controller and network). All the neutron agent are up and alive and the nova service is up too. I create network and upload images to glance.
When i try to launch an instance i get an error no valid host, Code 500. Find below the information in the log of nova-conductor
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager return func(*args, **kwargs)
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager File "/usr/lib/python2.7/site-packages/nova/scheduler/manager.py", line 169, in select_destinations
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager allocation_request_version, return_alternates)
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager File "/usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py", line 91, in select_destinations
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager allocation_request_version, return_alternates)
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager File "/usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py", line 256, in _schedule
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager claimed_instance_uuids)
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager File "/usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py", line 293, in _ensure_sufficient_hosts
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager raise exception.NoValidHost(reason=reason)
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager NoValidHost: H\xf4te non valide trouv\xe9. Le nombre d'h\xf4tes disponibles est insuffisant
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:49.961 47942 ERROR nova.conductor.manager
2019-04-24 20:08:50.651 47942 WARNING nova.scheduler.utils [req-930f1711-27d0-47af-9871-ca4dda3b5393 d162ebe5a0ab495fb4a863f52a858778 110bb0bc1bdb43478a76b16ed8bd7ce1 - default default] Failed to compute_task_build_instances: Hôte non valide trouvé. Le nombre d'hôtes disponibles est insuffisant
Traceback (most recent call last):
File "/usr/lib/python2.7/site-packages/oslo_messaging/rpc/server.py", line 226, in inner
return func(*args, **kwargs)
File "/usr/lib/python2.7/site-packages/nova/scheduler/manager.py", line 169, in select_destinations
allocation_request_version, return_alternates)
File "/usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py", line 91, in select_destinations
allocation_request_version, return_alternates)
File "/usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py", line 256, in _schedule
claimed_instance_uuids)
File "/usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py", line 293, in _ensure_sufficient_hosts
raise exception.NoValidHost(reason=reason)
NoValidHost: H\xf4te non valide trouv\xe9. Le nombre d'h\xf4tes disponibles est insuffisant
: NoValidHost_Remote: H\xf4te non valide trouv\xe9. Le nombre d'h\xf4tes disponibles est insuffisant
2019-04-24 20:08:50.655 47942 WARNING nova.scheduler.utils [req-930f1711-27d0-47af-9871-ca4dda3b5393 d162ebe5a0ab495fb4a863f52a858778 110bb0bc1bdb43478a76b16ed8bd7ce1 - default default] [instance: f9293645-5706-404a-ae79-849f4fa257e6] Setting instance to ERROR state.: NoValidHost_Remote: H\xf4te non valide trouv\xe9. Le nombre d'h\xf4tes disponibles est ...
Do you see the nova-compute service when running
openstack compute service list
?Is the compute node in the Placement database? You can install the Placement API
pip install osc-placement
then runopenstack resource provider list
. See https://docs.openstack.org/osc-placement.Also, if DEBUG logging is enabled, the nova-scheduler log should include information why no valid host was found.
If (speculation) you used packstack to install, and you accepted the default CentOS/RH partitioning scheme when installing the OS & the default packstack answers then most of the space is allocated to /home. & there's very little space for Glance/Cinder under / ==> this error (ask me how I know).
@Bernd Bausch hey, where can i enable that debug logging, which files and which lines? thanks
@brucema ..your speculation is right i use packstack and use the default Centos partitioning scheme, u seem to have experience on that issue, some advices to avoid it?
Debug logging: [DEFAULT] section in nova.conf. See https://docs.openstack.org/nova/lates....
@Patrick MAMIA I manually partitioned the disks during OS install and allocated the bulk of the space to / I also upped the size of
CONFIG_CINDER_VOLUMES_SIZE
in the packstack answer file.hey @brucema i'm trying to build another compute since, i got some issues to make it discover by the controller node so when i success i try your suggestions. Thanks