It looks like it's not an issue in nova nor cinder.
If you take the XML that nova is using (if you turn debug log, you'll see it in
nova-compute.log):
<disk type="network" device="disk">
<driver name="qemu" type="raw" cache="writeback"/>
<source protocol="rbd"
name="nova/volume-030ske6e-9348-4740-94dc-c8624b64ac0e">
<host name="192.168.225.9" port="6789"/>
<host name="192.168.225.10" port="6789"/>
<host name="192.168.225.23" port="6789"/>
</source>
<auth username="nova">
<secret type="ceph" uuid="6fccyte3-aa5c-4b5e-9627-88akwy82b131"/>
</auth>
<target bus="scsi" dev="sda"/>
<serial>025aa02e-9348-4740-94dc-c8624444ac0e</serial>
</disk>
And try to attach it to the domain manually:
# virsh attach-device instance-0000080a test-dev.xml --live
error: Failed to attach device from test-dev.xml
error: internal error: unable to execute QEMU command 'object-add': attempt to
add duplicate property 'scsi0-0-0-0-secret0' to object (type 'container')
You will see the same error. It's most likely a bug in libvirt, which in
CentOS 7 is at version 2.0.0.
Reported libvirt bug:
https://bugzilla.redhat.com/show_bug.cgi?id=1459690
** Bug watch added: Red Hat Bugzilla #1459690
https://bugzilla.redhat.com/show_bug.cgi?id=1459690
** Changed in: nova
Status: New => Invalid
** Changed in: nova
Status: Invalid => Confirmed
--
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1678694
Title:
Can't attach volume to volume-backed instance
Status in OpenStack Compute (nova):
Confirmed
Bug description:
When trying to attach a cinder volume to an instance that was launched
with a cinder root drive, the attachment silently fails both on the
CLI and in Horizon. From the nova-compute.log on the hypervisor,
"libvirtError: internal error: unable to execute QEMU command 'object-
add': attempt to add duplicate property 'scsi0-0-0-0-secret0' to
object (type 'container')" - so it seems like nova isn't aware of the
existing cinder root drive.
Steps to reproduce:
1) Launch an instance (either via CLI or in Horizon), specifying a cinder
volume as the root
2) Create a volume and try to attach it to the instance
Expected outcome:
Second volume attaches successfully to the instance
Actual outcome:
Volume silently fails to attach, remains in "available" state, error log as
above in nova-compute.log
Environment
CentOS 7.2
Mitaka (13.1.2-1.el7)
libvirt+kvm
ceph storage
To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1678694/+subscriptions
--
Mailing list: https://launchpad.net/~yahoo-eng-team
Post to : yahoo-eng-team@lists.launchpad.net
Unsubscribe : https://launchpad.net/~yahoo-eng-team
More help : https://help.launchpad.net/ListHelp
