Ask Your Question
1

RDO Icehouse with ceph - libvirt error

asked 2014-12-12 10:12:29 -0500

stevea gravatar image

updated 2014-12-12 10:53:10 -0500

smaffulli gravatar image

I've done a single RDO all in one install using Icehouse repos with neutron enabled all is ok with the default installation until I integrate ceph

I followed this guidance for ceph integration (I appreciate its havana rather than icehouse) https://openstack.redhat.com/Using_Ce...

I use non standard pool and usernames - so my glance is iceimages, cinder is icevolumes cinder-backup is icebackups

Glance is ok - populates images in ceph iceimages pool ok, can upload download so its all good

Cinder itself is ok, creates volumes, volume ID matches whats in the icevolumes pool so cinder is ok

The problem is with libvirt

I get this error when i try and launch a VM using any method (note - libvirt auth key redacted)

2014-12-12T15:49:24.719026Z qemu-kvm: -drive file=rbd:icevolumes/volume-4cf5a1fd-c725-4245-abc0-427d02ce5846:id=icevolumes:key=:auth_supported=cephx\;none:mon_host=10.80.0.64\:6789\;10.80.0.65\:6789\;10.80.0.66\:6789,if=none,id=drive-virtio-disk0,format=raw,serial=4cf5a1fd-c725-4245-abc0-427d02ce5846,cache=none: could not open disk image rbd:icevolumes/volume-4cf5a1fd-c725-4245-abc0-427d02ce5846:id=icevolumes:key=<removed>=:auth_supported=cephx\;none:mon_host=10.80.0.64\:6789\;10.80.0.65\:6789\;10.80.0.66\:6789: No such file or directory 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] Traceback (most recent call last): 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/compute/manager.py", line 1328, in _build_instance 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] set_access_ip=set_access_ip) 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/compute/manager.py", line 393, in decorated_function 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] return function(self, context, *args, **kwargs) 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/compute/manager.py", line 1740, in spawn 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] LOG.exception(('Instance failed to spawn'), instance=instance) 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/openstack/common/excutils.py", line 68, in exit 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] six.reraise(self.type_, self.value, self.tb) 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/compute/manager.py", line 1737, in _spawn 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] block_device_info) 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/virt/libvirt/driver.py", line 2294, in spawn 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] block_device_info) 2014-12-12 15:49:25.909 15892 TRACE nova.compute.manager [instance: 47f7b479-75c2-4ab3-bd00-52a4477a2dab] File "/usr/lib/python2.6/site-packages/nova/virt/libvirt/driver.py", line 3695, in ...
(more)
edit retag flag offensive close merge delete

Comments

Qemu looks to be the problem - it doesnt have rbd as a supported format. So it looks like the wrong version needed. Any ideas on how to set RDO up using inktank packages ?

stevea gravatar imagestevea ( 2014-12-15 11:51:58 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
0

answered 2014-12-16 10:00:32 -0500

stevea gravatar image

updated 2014-12-16 10:02:11 -0500

Solved: The core issue is the version of qemu installed by default with an icehouse RDO installation to resolve this, simply add in the 'ceph extras repo' when first building out an RDo instance as follows:

note this is tested on RHEL6.5 only

  1. sudo yum install http://rdo.fedorapeople.org/openstack...
  2. sudo yum install -y openstack-packstack
  3. configure ceph-extras repo as follows: nano /etc/yum.repos.d/ceph-extras.repo

[ceph-extras]

name=Ceph Extras baseurl=http://ceph.com/packages/ceph-extras/rpm/rhel6.5/$basearch enabled=1 priority=2 gpgcheck=1 type=rpm-md gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc

[ceph-qemu-source]

name=Ceph Extras Sources baseurl=http://ceph.com/packages/ceph-extras/rpm/rhel6.5/SRPMS enabled=1 priority=2 gpgcheck=1 type=rpm-md

save and close

  1. packstack --allinone --provision-all-in-one-ovs-bridge=n (im setting up Neutron, you may just want to run the --allinone config instead

install then completes successfully with rbd supported qemu in place

test qemu support as follows: qemu-img | grep "Supported format"

returns.... Supported formats: raw cow qcow vdi vmdk cloop dmg bochs vpc vvfat qcow2 qed vhdx parallels nbd blkdebug host_cdrom host_floppy host_device file gluster gluster gluster gluster rbd

note 'rbd' <<< this indicates ceph support is available and is the correct version needed

edit flag offensive delete link more

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

1 follower

Stats

Asked: 2014-12-12 10:12:29 -0500

Seen: 536 times

Last updated: Dec 16 '14