Confusing fault reasion when the flavors disk size was too small

[Attila Fazekas]

Fedora-x86_64-20-20140407-sda has 2 GiB virtual size.

$ nova boot fed_1G_2 --image Fedora-x86_64-20-20140407-sda --flavor 1 --key-name mykey
$ nova show fed_1G_2
+--------------------------------------+------------------------------------------------------------------------------------------+
| Property | Value |
+--------------------------------------+------------------------------------------------------------------------------------------+
| OS-DCF:diskConfig | MANUAL |
| OS-EXT-AZ:availability_zone | nova |
| OS-EXT-STS:power_state | 0 |
| OS-EXT-STS:task_state | - |
| OS-EXT-STS:vm_state | error |
| OS-SRV-USG:launched_at | - |
| OS-SRV-USG:terminated_at | - |
| accessIPv4 | |
| accessIPv6 | |
| config_drive | |
| created | 2014-06-17T07:35:43Z |
| fault | {"message": "No valid host was found. ", "code": 500, "created": "2014-06-17T07:35:44Z"} |
| flavor | m1.tiny (1) |
| hostId | a904a292f4eb7f6735bef786c4a240a0b9240a6bc4f002519cb0e2b7 |
| id | 3c908a54-9682-40ad-8f12-a5bf64066660 |
| image | Fedora-x86_64-20-20140407-sda (085610a8-77ae-4bc8-9a28-3bcc1020e06e) |
| key_name | mykey |
| metadata | {} |
| name | fed_1G_2 |
| os-extended-volumes:volumes_attached | [] |
| private network | 10.1.0.5 |
| security_groups | default |
| status | ERROR |
| tenant_id | 1d26ad7003cf47e5b0107313be4832c3 |
| updated | 2014-06-17T07:35:44Z |
| user_id | bf52e56b9ca14648b391c5b6d490a0c1 |
+--------------------------------------+------------------------------------------------------------------------------------------+

$ # nova flavor-list
+-----+-----------+-----------+------+-----------+---------+-------+-------------+-----------+
| ID | Name | Memory_MB | Disk | Ephemeral | Swap_MB | VCPUs | RXTX_Factor | Is_Public |
+-----+-----------+-----------+------+-----------+---------+-------+-------------+-----------+
| 1 | m1.tiny | 512 | 1 | 0 | | 1 | 1.0 | True |
| 2 | m1.small | 2048 | 20 | 0 | | 1 | 1.0 | True |
| 3 | m1.medium | 4096 | 40 | 0 | | 2 | 1.0 | True |
| 4 | m1.large | 8192 | 80 | 0 | | 4 | 1.0 | True |
| 42 | m1.nano | 64 | 0 | 0 | | 1 | 1.0 | True |
| 451 | m1.heat | 1024 | 0 | 0 | | 2 | 1.0 | True |
| 5 | m1.xlarge | 16384 | 160 | 0 | | 8 | 1.0 | True |
| 84 | m1.micro | 128 | 0 | 0 | | 1 | 1.0 | True |
+-----+-----------+-----------+------+-----------+---------+-------+-------------+-----------+

Many images requires minimum 2,5,10 Gib as minimum disk size, the 1 used by m1.tiny is frequently not enough.
It might be increased to 10 or the original 0 should be restored.

This bug is about why I see 'message": "No valid host was found. ", "code": 500, "created": "2014-06-17T07:35:44Z"'
when "Flavor's disk is too small for requested image'" was raised on the n-cpu side.
It is confusing, 'No valid host was found' type of messages sounds like there is no n-cpu running, or all of them full.

instance: f62b56da-d1fa-4dc2-ae37-42b8fde3d3a5] Instance failed to spawn
 Traceback (most recent call last):
   File "/opt/stack/new/nova/nova/compute/manager.py", line 2064, in _build_resources
     yield resources
   File "/opt/stack/new/nova/nova/compute/manager.py", line 1966, in _build_and_run_instance
     block_device_info=block_device_info)
   File "/opt/stack/new/nova/nova/virt/libvirt/driver.py", line 2233, in spawn
     admin_pass=admin_password)
   File "/opt/stack/new/nova/nova/virt/libvirt/driver.py", line 2607, in _create_image
     project_id=instance['project_id'])
   File "/opt/stack/new/nova/nova/virt/libvirt/imagebackend.py", line 182, in cache
     *args, **kwargs)
   File "/opt/stack/new/nova/nova/virt/libvirt/imagebackend.py", line 374, in create_image
     prepare_template(target=base, max_size=size, *args, **kwargs)
   File "/opt/stack/new/nova/nova/openstack/common/lockutils.py", line 249, in inner
     return f(*args, **kwargs)
   File "/opt/stack/new/nova/nova/virt/libvirt/imagebackend.py", line 172, in fetch_func_sync
     fetch_func(target=target, *args, **kwargs)
   File "/opt/stack/new/nova/nova/virt/libvirt/utils.py", line 658, in fetch_image
     max_size=max_size)
   File "/opt/stack/new/nova/nova/virt/images.py", line 110, in fetch_to_raw
     raise exception.FlavorDiskTooSmall()
 FlavorDiskTooSmall: Flavor's disk is too small for requested image.

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

Here's a theory:

We should catch the 3 Exceptions from _check_requested_image [1] in _build_and_run_instance [2] and NOT throw RescheduledException. Currently the FlavorDiskTooSmall ends up in the last Exception block which reschedules the deploy a few times and finally ends up throwing NoValidHost.

[1] https://github.com/openstack/nova/blob/master/nova/compute/api.py#L598
[2] https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L1999

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/101090

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/101090
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=da9891a30b3a3eac39458d829abdd5bc50cf876e
Submitter: Jenkins
Branch: master

commit da9891a30b3a3eac39458d829abdd5bc50cf876e
Author: Thang Pham <email address hidden>
Date: Thu Jun 19 00:19:13 2014 -0400

    Catch image and flavor exceptions in _build_and_run_instance

    If a user selects a flavor with a disk size that is smaller than
    the requested image, a FlavorDiskTooSmall exception is thrown.
    However, this exception is not specifically caught in
    _build_and_run_instance, but is caught as generic exception,
    causing a RescheduledException to be thrown. This forces the
    instance to be rescheduled a few times before finally throwing
    a NoValidHost exception. It causes confusion because the actual
    cause of it is that the disk size is smaller than the requested
    image. The following patch properly catches exceptions thrown by
    _check_requested_image in _build_and_run_instance.

    Change-Id: Ic8b435ba7e8721aa8ca54f29c1ad808833a7608d
    Closes-Bug: #1330856