Ask Your Question
0

VM creation fail : No vaild host was found

asked 2016-11-10 22:30:40 -0600

openstackstarter gravatar image

Hi. I'm deploying Openstack Mitaka with 1 controller and 1networknode and 3compute node. Everything seems to work fine, and I'm able to launch instances also. But now I'm trying to launch instance with large capacity options, like m1.xlarge, and it fails, saying No valid host was found. I found out that Filters filtering out all my compute nodes as a valid host. And I believe overcommit isn't really a way to get this done, since it might crash when disk or ram of that node are fully used. So my questions..

  1. Do I have to assign RAM or diskstorage constrained to the amount of the maximum of one compute node? If so, please tell me the reason? I'm confused.. isn't that against the concept of cloud?
  2. my hypervisor-stats shows all my three compute-nodes, but filters are not counting compute-node2 as you can see in the error log. What configurations or files should I be looking at?

tail /var/log/nova/nova-scheduler.log - diskfilter

 2016-11-10 18:53:12.288 10845 DEBUG nova.scheduler.host_manager [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Update host state with aggregates: [] _locked_update /usr/lib/python2.7/dist-packages/nova/scheduler/host_manager.py:171
    2016-11-10 18:53:12.288 10845 DEBUG nova.scheduler.host_manager [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Update host state with service dict: {'binary': u'nova-compute', 'deleted': False, 'created_at': datetime.datetime(2016, 10, 15, 7, 31, 22, tzinfo=<iso8601.Utc>), 'updated_at': datetime.datetime(2016, 11, 10, 9, 53, 11, tzinfo=<iso8601.Utc>), 'report_count': 12760, 'topic': u'compute', 'host': u'compute-node3', 'version': 9, 'disabled': False, 'forced_down': False, 'last_seen_up': datetime.datetime(2016, 11, 10, 9, 53, 11, tzinfo=<iso8601.Utc>), 'deleted_at': None, 'disabled_reason': None, 'id': 8} _locked_update /usr/lib/python2.7/dist-packages/nova/scheduler/host_manager.py:174
    2016-11-10 18:53:12.289 10845 DEBUG nova.scheduler.host_manager [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Update host state with instances: {} _locked_update /usr/lib/python2.7/dist-packages/nova/scheduler/host_manager.py:177
    2016-11-10 18:53:12.289 10845 DEBUG oslo_concurrency.lockutils [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Lock "(u'compute-node3', u'compute-node3')" released by "nova.scheduler.host_manager._locked_update" :: held 0.003s inner /usr/lib/python2.7/dist-packages/oslo_concurrency/lockutils.py:285
    2016-11-10 18:53:12.290 10845 WARNING nova.scheduler.host_manager [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] No compute service record found for host compute-node2
    2016-11-10 18:53:12.290 10845 DEBUG nova.filters [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Starting with 2 host(s) get_filtered_objects /usr/lib/python2.7/dist-packages/nova/filters.py:70
    2016-11-10 18:53:12.291 10845 DEBUG nova.filters [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Filter RetryFilter returned 2 host(s) get_filtered_objects /usr/lib/python2.7/dist-packages/nova/filters.py:104
    2016-11-10 18:53:12.291 10845 DEBUG nova.filters [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Filter AvailabilityZoneFilter returned 2 host(s) get_filtered_objects /usr/lib/python2.7/dist-packages/nova/filters.py:104
    2016-11-10 18:53:12.292 10845 DEBUG nova.filters [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 fda14487a68f436da7b55fe218798ea2 - - -] Filter RamFilter returned 2 host(s) get_filtered_objects /usr/lib/python2.7/dist-packages/nova/filters.py:104
    2016-11-10 18:53:12.292 10845 DEBUG nova.scheduler.filters.disk_filter [req-10c398e2-6611-4884-8883-cdd4c2445848 6abfef77e9984fc58bb5decc566d12c2 ...
(more)
edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
2

answered 2016-11-11 05:50:54 -0600

volenbovsky gravatar image

Hi, 1. you can't aggregate RAM (or CPU) from several Compute Hosts. There are no known SW/hypervisors/VM managers/anything that will allow you to have RAM or CPU as resource pooled/shared so that one VM can use such resources from several different physical/virtual servers.

Disaggregation of CPU/RAM will come with things Intel RackScale Architecture but in that case actually OpenStack (and eg. KVM/QEMU) is not too much in focus according to my understanding

One way to think about that the CPU/RAM of your OS/processes should be really in 'one physical place', and even having it 'slightly in other place like other socket will have penalty (see NUMA), but having that 'span' several servers require a whola different bandwidth

  1. Apparently that is ' No compute service record found for host compute-node2' from log See nova service-list and see if nova-compute is there...
edit flag offensive delete link more

Comments

Thank you fore your answer! This really helped me grasp better understanding on openstack cloud. So now I get that what you mean resources should be really in one physical place. On second question, nova-compute and service list shows compute-node2 just fine.. still working on it :-<

openstackstarter gravatar imageopenstackstarter ( 2016-11-13 03:22:02 -0600 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

2 followers

Stats

Asked: 2016-11-10 22:30:40 -0600

Seen: 174 times

Last updated: Nov 11 '16