Usage of live migration feature(nova是支持在线迁移的，其他链接)

Environments

 

• OS: Ubuntu lucid 10.04 for both instances and host.  
• NFS: nova-install-dir/instances has to be mounted by shared storage.( this version is tested using NFS)  
• Instances : Instance must keep running without any EBS volume.   

 

Pre-requisite settings

 

(a) shared storage As mentioned above, shared storage is inevitable for the live_migration functionality. An example is NFS( my test environment ), and example setting is described below. Prepare NFS server machine( nova-api server is OK), and add below line /etc/exports:   

  

 nova-install-dir/instances a.b.c.d/255.255.0.0(rw,sync,fsid=0,no_root_squash

 

where "nova-install-dir" is the directory which openstack is installed, and add appropriate ip address and netmask for "a.b.c.d/255.255.0.0" , which should include compute nodes mounting this directory. Then restart nfs server.   

   

 /etc/init.d/nfs-kernel-server restart
 /etc/init.d/idmapd restart

 

Also, at any compute nodes, add below line to /etc/fstab:   

  

 a.b.c.d:/ DIR nfs4 defaults 0 0

 

where a.b.c.d is nfs server's ipaddress and "DIR" must be same as 'instances_path'( seenova.compute.manager for the default value)   

Then try to mount at compute node. Check exported directory is successfully mounted.   

  

 mount -a -v

 

If fail, try this at any hosts.

iptables -F

Also, check file/daemon permissions. We expect any nova daemons are running as root.

root@openstack2-api:/opt/nova-2010.4# ps -ef | grep nova
root 5948 5904 9 11:29 pts/4 00:00:00 python /opt/nova-2010.4//bin/nova-api
root 5952 5908 6 11:29 pts/5 00:00:00 python /opt/nova-2010.4//bin/nova-objectstore
... (snip)

"instances/" directory can be seen from server side:

root@openstack:~# ls -ld nova-install-dir/instances/
drwxr-xr-x 2 root root 4096 2010-12-07 14:34 nova-install-dir/instances/

Also, client(compute node) side:

# ls -ld nova-install-dir/instances/
drwxr-xr-x 2 root root 4096 2010-12-07 14:34 nova-install-dir/instances/

(b) libvirt settings In default configuration, this feature use simple tcp protocol(qemu+tcp://).To use this protocol, below configuration is necessary.

Modify /etc/libvirt/libvirt.conf :

before : #listen_tls = 0
after : listen_tls = 0

before : #listen_tcp = 1
after : listen_tcp = 1

Adding following line to /etc/libvirt/libvirt.conf :

auth_tcp = "none"

Modify /etc/init/libvirt-bin.conf

 before : exec /usr/sbin/libvirtd -d
 after : exec /usr/sbin/libvirtd -d -l

Modify modify /etc/default/libvirt-bin

before :libvirtd_opts=" -d"
after :libvirtd_opts=" -d -l"

then, restart libvirt

stop libvirt-bin && start libvirt-bin ps -ef | grep libvirt

Make sure you get the below result.

root@openstack2:/opt/nova-2010.2# ps -ef | grep libvirt
root 1145 1 0 Nov27 ? 00:00:03 /usr/sbin/libvirtd -d -l

if you would like to use qemu+ssh or other protocol, change "live_migration_uri" flag. by adding "--live_migration_uri" to /etc/nova/nova.conf and make appropriate settings described at http://libvirt.org/.

 

Usage

• To get a list of physical hosts,
  • nova-manage host list
• To get a available pysical resource of each host,
  • nova-manage host shows hostname
  • An example result is below. The 1st line shows total amount of resource that the specified host has. The 2nd and latter lines show usage resource per project. This command is created because admins can decide which host should be a destination of live migration.

  • HOST PROJECT cpu mem(mb) disk(gb)
    openstack2-c2 16 32232 878 > openstack2-c2 admin 1 2048 20

• For live migration,
  • nova-manage instances live_migration ec2-id(i-xxxx) destination-host-name.
  • Once this command is executed, admins will check the status through euca-describe-instances. The status is changed from 'running' to 'migrating', and changed to 'running' when live migration finishes. Note that it depends on an environment how long it takes to live migration finishes. If it finishes too fast, admins cannot see 'migrating' status.

  • # euca-describe-instances
    Reservation:r-qlg3favp
    RESERVATION r-qlg3favp adminINSTANCE i-2ah453 ami-tiny 12.1.0.14 10.0.0.3
    migrating testkey (admin, openstack2-c2) 0 m1.small 2010-11-28 16:09:16

  • When live migration finishes successfully, admin can check the part of euca-describe-instances which shows physical node information. ( only when euca-describe-instances is executed by admin user ) Admins also can check live migration source compute node logfile which may show a log.

  • Live migration i-xxxx to DESTHOST finishes successfully.

 

Trouble shooting

When live migration fails somehow, error messages are shown at:

• scheduler logfile
• source compute node logfile
• dest compute node logfile

在线迁移：http://wiki.openstack.org/UsageOfLiveMigration

CentOS安装：http://wiki.openstack.org/NovaInstall/CentOSNotes

NovaCategory：http://wiki.openstack.org/CategoryNova

Nova与Eucalyptus的对比：http://wiki.openstack.org/Nova/EucalyptusFeatureComparison?highlight=%28\bCategoryNova\b%29

Nova综述：http://wiki.openstack.org/Nova?highlight=%28\bCategoryNova\b%29