1.4.5. Start and stop

Once the fast path configuration is complete, we can start Virtual Accelerator.

First, make sure that the Network Manager service is stopped and disabled:

# systemctl stop NetworkManager.service
# systemctl disable NetworkManager.service

Then, start Virtual Accelerator as follows:

# systemctl start virtual-accelerator.target

Note

Please remember that the fast path takes control of all supported physical network ports by default. You might want to prevent the fast path from taking your management interface, using the configuration wizard.

By default, since Virtual Accelerator version 1.9.3, re-connection with existing VMs, managed by the previous Virtual Accelerator instance, is automatic. If this behavior is not expected, the previous Virtual Accelerator state must be deleted before the start operation.

# fp-vdev clean fp-vdev-conf
# systemctl start virtual-accelerator.target

To stop it:

# systemctl stop virtual-accelerator.target

Note

If stop operation is done to update some fast path parameters, existing VMs can be kept. For that VMs information are automatically saved before the stop and restored after the start.

# service virtual-accelerator stop
# # update system configuration
# service virtual-accelerator start

No change can be done on VMs between the save and restore operations.

To start Virtual Accelerator at boot time:

# systemctl enable virtual-accelerator.target

Note

When the Virtual Accelerator service is started or stopped, it will attempt to restart all interfaces configured in the system by calling ifdown -a followed by ifup -a.

Note

libvirt must be restarted after the first Virtual Accelerator start to actually see the hugepages allocated in VM_MEMORY.

Note

If your setup comprises Open vSwitch bridges, the Open vSwitch service must be restarted each time Virtual Accelerator is restarted.