Booting instance with flavor too small returns NoValidHost to client when error should be e.g. InstanceTypeDiskTooSmall

[Kevin Jackson]

Ubuntu 12.04 with Havana (precise-updates and precise-proposed)

ii nova-ajax-console-proxy 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - AJAX console proxy - transitional package
ii nova-api 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - API frontend
ii nova-cert 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - certificate management
ii nova-common 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - common files
ii nova-conductor 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - conductor service
ii nova-consoleauth 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - Console Authenticator
ii nova-doc 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - documentation
ii nova-novncproxy 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - NoVNC proxy
ii nova-objectstore 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - object store
ii nova-scheduler 1:2013.2-0ubuntu1~cloud0 OpenStack Compute - virtual machine scheduler
ii python-nova 1:2013.2-0ubuntu1~cloud0 OpenStack Compute Python libraries
ii python-novaclient 1:2.15.0-0ubuntu1~cloud0 client library for OpenStack Compute API

When launching an instance of a type that is too small, the error returned to the client is NoValidHost when the real error message is (in this case) "InstanceTypeDiskTooSmall: Instance type's disk is too small for requested image" - this is confusing to the user and sends you on a different route of troubleshooting.

# nova show test1

fault | {u'message': u'No valid host was found. Exceeded max scheduling attempts 3 for instance 75ed5622-dece-4c5a-9897-daa4fda1d829', u'code': 500, u'details': u' File "/usr/lib/python2.7/dist-packages/nova/scheduler/manager.py", line 147, in run_instance |
| | legacy_bdm_in_spec) |
| | File "/usr/lib/python2.7/dist-packages/nova/scheduler/filter_scheduler.py", line 85, in schedule_run_instance |
| | filter_properties, instance_uuids) |
| | File "/usr/lib/python2.7/dist-packages/nova/scheduler/filter_scheduler.py", line 306, in _schedule |
| | self._populate_retry(filter_properties, properties) |
| | File "/usr/lib/python2.7/dist-packages/nova/scheduler/filter_scheduler.py", line 275, in _populate_retry |
| | raise exception.NoValidHost(reason=msg) |
| | ', u'created': u'2013-10-27T20:16:07Z'}

Log entry in /var/log/nova/nova-scheduler.log

2013-10-27 20:16:03.731 11409 ERROR nova.scheduler.filter_scheduler [req-cabdc8cd-2a7e-439f-a112-3f91a890f530 108bf3466c8a4388974ecd1ce6fb005b 120926d0d0324d21a535ad870f6eb662] [instance: 75ed5622-dece-4c5a-9897-daa4fda1d829] Error from last host: compute (node compute.book): [u'Traceback (most recent call last):\n', u' File "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 1037, in _build_instance\n set_access_ip=set_access_ip)\n', u' File "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 1410, in _spawn\n LOG.exception(_(\'Instance failed to spawn\'), instance=instance)\n', u' File "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 1407, in _spawn\n block_device_info)\n', u' File "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/driver.py", line 2063, in spawn\n admin_pass=admin_password)\n', u' File "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/driver.py", line 2353, in _create_image\n project_id=instance[\'project_id\'])\n', u' File "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/imagebackend.py", line 174, in cache\n *args, **kwargs)\n', u' File "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/imagebackend.py", line 312, in create_image\n raise exception.InstanceTypeDiskTooSmall()\n', u"InstanceTypeDiskTooSmall: Instance type's disk is too small for requested image.\n"]

[Davanum Srinivas (DIMS) (dims-v)]

Guessing, this other bug https://bugs.launchpad.net/nova/+bug/1244316 is because of this same problem.

[Kevin Jackson (kevin-linuxservices)]

Marked as dupe.

[Guangya Liu (Jay Lau) (jay-lau-513)]

This is also a duplicate bug with 1161661.

The root cause is retryFilter was enabled by default and default value of max_attempts is 3, so after nova compute try 3 times, nova scheduler will report "no valid host" and this will be stored to table of instance_fault for the VM.

So when you run "nova show", it will show scheduler error "no valid host" but not nova compute error.