instance status turn to ERROR when running instance suspend

[Yogev Rabl]

Description of problem:
When trying to suspend an instance the instance's status turn to Error.
The instance's flavor details are:

+----------------------------+--------------------------------------+
| Property | Value |
+----------------------------+--------------------------------------+
| name | m1.small |
| ram | 2048 |
| OS-FLV-DISABLED:disabled | False |
| vcpus | 1 |
| extra_specs | {} |
| swap | |
| os-flavor-access:is_public | True |
| rxtx_factor | 1.0 |
| OS-FLV-EXT-DATA:ephemeral | 40 |
| disk | 20 |
| id | 7427e83a-5f96-43af-936b-a054191482ab |
+----------------------------+--------------------------------------+

Version-Release number of selected component (if applicable):

openstack-nova-common-2013.2.3-6.el6ost.noarch
openstack-nova-console-2013.2.3-6.el6ost.noarch
openstack-nova-network-2013.2.3-6.el6ost.noarch
python-novaclient-2.15.0-4.el6ost.noarch
python-nova-2013.2.3-6.el6ost.noarch
openstack-nova-compute-2013.2.3-6.el6ost.noarch
openstack-nova-conductor-2013.2.3-6.el6ost.noarch
openstack-nova-novncproxy-2013.2.3-6.el6ost.noarch
openstack-nova-scheduler-2013.2.3-6.el6ost.noarch
openstack-nova-api-2013.2.3-6.el6ost.noarch
openstack-nova-cert-2013.2.3-6.el6ost.noarch

How reproducible:
100%

Steps to Reproduce:
1. launch an instance from an iso image with the flavor as it is detailed above.
2. suspend the instance.

Actual results:
The instance status turns to ERROR.

Expected results:
The instance should be suspend

Additional info:
The error from the compute log is attached.

[Thang Pham (thang-pham)]

Have you tried using just virsh suspend <domain>? I wonder if it is just a libvirt bug because I have also noticed it.

[Yogev Rabl (yrabl)]

Thank you for the reply, I have tried it and the virsh suspend/resume worked properly. Is there anything else I can check?

[Thang Pham (thang-pham)]

What does /var/log/libvirt/libvirtd.log show? Are there error messages associated witht he suspend/resume?

[Yogev Rabl (yrabl)]

I've checked the libvirt log and there was no error during the time of the error in the compute log.

[Thang Pham (thang-pham)]

That is odd. The nova-compute log says it is a libvirt error:
    libvirtError: Requested operation is not valid: cannot do managed save for transient domain

[Nikola Đipanov (ndipanov)]

Copying the discussion from the original bug tracker for completeness - by danpb:

"""
There's not really enough info here to give a good idea of where the bug might lie. To start with we'd really need to capture some verbose debug logs of libvirt APIs that Nova is running. eg set these env vars for nova compute

 LIBVIRT_LOG_FILTERS="1:libvirt 1:qemu" LIBVIRT_LOG_OUTPUTS="1:file:/var/log/nova/libvirt-client.log"

With that info we'd be able to see if Nova was issuing a virDomainUndefine call to cause this situation, or whether libvirt is loosing track of the defined domain.
"""

So we'd need more info from the reporter to rule out the libvirt bug.

[Sean Dague (sdague)]

4 months in incomplete status, closing