Ask Your Question

Revision history [back]

I think currently openstack has no mechanism to detect host failure and automatically migrate all running VMs to other healthy hosts. This is a point where third parties doing business(They provide HA mechanism to openstack for legacy and non-cloud ready applications(stateful apps)). Live migration(with shared storage) is a good method for compute node maintenance with zero down time for all services(VMs) running on the compute server. But is not a way to provide service high availability and automatic fail-over in the case of host failure. To have succesful live migration, both source and destination hypervisors must be on running(healthy) state.

In general, OpenStack tries to target applications that support recovery and resiliency in the application itself. This is a new paradigm on the application development(Stateless cloud-ready application development).

(You can configure openstack to work with vmware vcenter to provide automatic VM recovery in the case of host failure. But this is propertary solution as I said before)

About the networking between VMs when migrated to other hosts due to live migration: if you setup and confgure your openstack networking service properly with right configuration values, no networking problems will be in the case of VM migration. But some limitation exists(e.g. your both compute nodes needs to be in the same network/subnet)

I think currently openstack has no mechanism to detect host failure and automatically migrate all running VMs to other healthy hosts. This is a point where third parties doing business(They provide HA mechanism to openstack for legacy and non-cloud ready applications(stateful apps)). apps)).

Live migration(with shared storage) is a good method for compute node maintenance with zero down time for all services(VMs) running on the compute server. But is not a way to provide service high availability and automatic fail-over in the case of host failure. To have succesful live migration, both source and destination hypervisors must be on running(healthy) state.

In general, OpenStack tries to target applications that support recovery and resiliency in the application itself. This is a new paradigm on the application development(Stateless cloud-ready application development).

(You can configure openstack to work with vmware vcenter to provide automatic VM recovery in the case of host failure. But this is propertary solution as I said before)

About the networking between VMs when migrated to other hosts due to live migration: if you setup and confgure your openstack networking service properly with right configuration values, no networking problems will be in the case of VM migration. But some limitation exists(e.g. your both compute nodes needs to be in the same network/subnet)

I think currently openstack has no mechanism to detect host failure and automatically migrate all running VMs to other healthy hosts. This is a point where third parties doing business(They provide HA mechanism to openstack for legacy and non-cloud ready applications(stateful apps)).

Live migration(with shared storage) is a good method for compute node maintenance with zero down time for all services(VMs) running on the compute server. But is not a way to provide service high availability and automatic fail-over in the case of host failure. To have succesful live migration, both source and destination hypervisors must be on running(healthy) state.

In general, OpenStack tries to target applications that support recovery and resiliency in the application itself. This is a new paradigm on the application development(Stateless cloud-ready application development).

(You can configure openstack to work with vmware vcenter to provide automatic VM recovery in the case of host failure. But this is propertary solution as I said before)

About the networking between VMs when migrated to other hosts due to live migration: if you setup and confgure your openstack networking service properly with right configuration values, no networking problems will be in the case of VM migration. But some limitation exists(e.g. your both compute nodes needs need to be in the same network/subnet)

You can use resume_guests_state_on_host_boot=true value on your nova.conf to force nova resume your VMs when the host machine become available after restarting.

But about the automatic VM migration: I think currently openstack has no mechanism to detect host failure and automatically migrate all running VMs to other healthy hosts. This is a point where third parties doing business(They provide HA mechanism to openstack for legacy and non-cloud ready applications(stateful apps)).

Live migration(with shared storage) is a good method for compute node maintenance with zero down time for all services(VMs) running on the compute server. But is not a way to provide service high availability and automatic fail-over in the case of host failure. To have succesful live migration, both source and destination hypervisors must be on running(healthy) state.

In general, OpenStack tries to target applications that support recovery and resiliency in the application itself. This is a new paradigm on the application development(Stateless cloud-ready application development).

(You can configure openstack to work with vmware vcenter to provide automatic VM recovery in the case of host failure. But this is propertary solution as I said before)

About the networking between VMs when migrated to other hosts due to live migration: if you setup and confgure your openstack networking service properly with right configuration values, no networking problems will be in the case of VM migration. But some limitation exists(e.g. your both compute nodes need to be in the same network/subnet)

You can use resume_guests_state_on_host_boot=true value on your nova.conf to force nova resume your VMs when the host machine become becomes available after restarting.

But about the automatic VM migration: I think currently openstack has no mechanism to detect host failure and automatically migrate automatic migration of all running VMs to other healthy hosts. This is a point where third parties doing business(They provide HA mechanism to openstack for legacy and non-cloud ready applications(stateful apps)).

Live migration(with shared storage) is a good method for compute node maintenance with zero down time for all services(VMs) running on the compute server. But is not a way to provide service high availability and automatic fail-over in the case of host failure. To have succesful live migration, both source and destination hypervisors must be on running(healthy) state.

In general, OpenStack tries to target applications that support recovery and resiliency in the application itself. This is a new paradigm on the application development(Stateless cloud-ready application development).

(You can configure openstack to work with vmware vcenter to provide automatic VM recovery in the case of host failure. But this is propertary solution as I said before)

About the networking between VMs when migrated to other hosts due to live migration: if you setup and confgure your openstack networking service properly with right configuration values, no networking problems will be in the case of VM migration. But some limitation exists(e.g. your both compute nodes need to be in the same network/subnet)