nova live-migrartion failed



I'm trying to test the nova live-migration, but i have always this output:
ERROR: Live migration of instance b6607c9b-cedb-4749-9366-68256f922171 to host controller-node failed (HTTP 400) (Request-ID: req-59d7e5a4-1639-40aa-9328-b04212338208)

here's my services:
Binary Host Zone Status State Updated_At
nova-consoleauth controller-node internal enabled :-) 2013-07-08 08:46:30
nova-cert controller-node internal enabled :-) 2013-07-08 08:46:30
nova-conductor controller-node internal enabled :-) 2013-07-08 08:46:30
nova-scheduler controller-node internal enabled :-) 2013-07-08 08:46:30
nova-compute compute-node lannion enabled :-) 2013-07-08 08:46:31
nova-compute controller-node lannion enabled :-) 2013-07-08 08:46:30

Here's the VM i'm tyring to migrate from the compute to the controller-node,
b6607c9b-cedb-4749-9366-68256f922171 | precise-admin | MIGRATING | priv1=10.10.x.x, 172.25.x.x |

It was active before the execution of the command, and now it is always on the migrarting status

Question information

English Edit question
OpenStack Compute (nova) Edit question
No assignee Edit question
Solved by:
Last query:
Last reply:
Revision history for this message
NAKKOUZI (nour-nakouzi) said :

I installed the nfs4server on the controller-node, and the nfs4client on both compute-nodes, the I mount a shared folder on /var/lib/nova/instances, what i noticed is, after creation the instance, the owner is root:root in /var/lib/nova/instances, and not nova:nova, so if i unmount the folder i can create instances, else the VM is on the spawming status

Any Ideas

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

If i create a file or a folder in this mounted point, i have the nobody:nogroup the user and the group that belongsto that file

Revision history for this message
Xiang Hui (xianghui) said :

Did you test the live-migration with shared storage? did you set 'instances_shared_storage = true' and 'instance_path=$shared' in the conf, please put the config files here.

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

Here's My nova.conf file an both compute-nodes:

state_path = /var/lib/nova
lock_path = /var/lock/nova
root_helper = sudo nova-rootwrap /etc/nova/rootwrap.conf
api_paste_config = /etc/nova/api-paste.ini
enabled_apis = ec2,osapi_compute,metadata
sql_connection = mysql://nova:nova@192.168.x.x/nova
default_availability_zone = lannion


# Logging
verbose = False
debug = False
logdir = /var/log/nova

# Identity
auth_strategy = keystone

# Volumes
volume_api_class = nova.volume.cinder.API

# Images
glance_api_servers = 192.168.x.x:9292

# Compute
compute_driver = libvirt.LibvirtDriver
libvirt_type = kvm
libvirt_inject_key = True
libvirt_inject_partition = -1
libvirt_wait_soft_reboot_seconds = 300
libvirt_snapshot_compression = True
instance_build_timeout = 900
reboot_timeout = 300
instance_usage_audit = False
instance_usage_audit_period = day
resume_guests_state_on_host_boot = True
instance_name_template = instance-%08x

# Security groups
firewall_driver = nova.virt.firewall.NoopFirewallDriver
security_group_api = quantum

# Networking
my_ip = 192.168.x.x
network_api_class =
quantum_admin_auth_url = http://192.168.x.x:35357/v2.0/
quantum_admin_tenant_name = service
quantum_admin_username = quantum
quantum_admin_password = servicepass
quantum_auth_strategy = keystone
quantum_url = http://192.168.x.x:9696/
force_dhcp_release = True

# Metadata
metadata_host = 192.168.x.x
metadata_port = 8775
service_quantum_metadata_proxy = True
#quantum_metadata_proxy_shared_secret =

# Policy
policy_file = policy.json
allow_instance_snapshots = True
allow_migrate_to_same_host = False
allow_resize_to_same_host = False

# Messaging
rpc_backend = nova.openstack.common.rpc.impl_kombu
rabbit_host = 192.168.x.x
# rabbit_port = 5672
# rabbit_userid = rabbit
# rabbit_password = rabbit
# rabbit_virtual_host = /

# Scheduler
max_instances_per_host = 50
reserved_host_disk_mb = 10240
reserved_host_memory_mb = 1024

# Live migration
live_migration_bandwidth = 0

vnc_enabled = True
vnc_port = 5900
vnc_port_total = 10000
vnc_password = None
novncproxy_base_url = http://172.25.x.x:6080/vnc_auto.html
vncserver_listen =
vncserver_proxyclient_address = 192.168.x;x
vnc_keymap = en-us

# EC2
ec2_host = 192.168.x.x
ec2_port = 8773
ec2_private_dns_show_ip = True
keystone_ec2_url = http://192.168.x.x:5000/v2.0/ec2tokens

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

Any Ideas.

Revision history for this message
Xiang Hui (xianghui) said :

If you are using nfs, the live-migration with shared storage is active,

make sure you have prepare the env referred by

1. check NOVA-INST-DIR/instances are same on all your hosts
2. run 'chown nova:nova NOVA-INST-DIR/instances' on all your hosts to solve the user group problem.

If still got some problems, up the erros.

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

I'm trying to change ther permission on the nodes:
chown nova:nova /var/lib/nova/instances

I have always this problem
chown: changing ownership of `/var/lib/nova/instances': Operation not permitted

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

In the doc, they said that:
Ensure that the UID and GID of your nova and libvirt users are identical between each of your servers. This ensures that the permissions on the NFS mount will work correctly.

I have different UID and GID for nova and libvirt in all the nodes, i change that manually, i chose a number

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

If I look to the schedular log, i ave the following
2013-07-10 10:00:20.927 5938 TRACE nova.openstack.common.rpc.amqp MigrationError: Migration error: Unable to migrate c40e3045-05b3-4c03-94f8-b37d3ed87e02 to compute-node-2: Lack of memory

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

So i added memory to the compute-node-2, now when executing the command nova live-migration 4dba73a9-5ab5-4202-8254-05b641abd92d compute-node-2, i have nothing on output

but the VM is always on the compute-node not the compute-node-2, so the migration is failed ??

Revision history for this message
Xiang Hui (xianghui) said :

set debug=True on /etc/nova/nova.conf to track the error, then up the log.

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

When i check the nova-compute log, i have this line:
2013-07-10 10:27:03.186 1627 ERROR nova.virt.libvirt.driver [-] [instance: 4dba73a9-5ab5-4202-8254-05b641abd92d] Live Migration failure: operation failed: Failed to connect to remote libvirt URI qemu+tcp://compute-node-2/system

Revision history for this message
NAKKOUZI (nour-nakouzi) said :

Finally i found the problem, i add the ip address of compute-node-2 in /etc/hosts