Related tests for OpenStack (JUNO) neutron

This two days to test the company's OpenStack environment, because the environment is built by other colleagues, my side is only responsible for the relevant tests, a single server installation

Controller node, and the other two servers have network Node+compute node installed. The approximate environment is as follows:

The Outer network IP network segment is 192.168.1.0/24, wherein 192.168.1.100~192.168.1.200 is the FLOATINGIP distribution range, the intranet IP network segment is 172.16.1.0/24.

On the compute0 there is a instance VM00, assigned the external network IP192.168.1.102, the intranet IP is 172.16.1.2. There's a instance on compute1.

VM01, assigned the external network IP192.168.1.103, the intranet IP is 172.16.1.3. VM00 and VM01 are on the same intranet subnet.

Intranet and extranet are interconnected through router, so they can communicate with VM00 and VM01 on controller.

Now you need to remove the compute0, you need to migrate the VM00 to Compute1, and then close the compute0. Using Nova migrate and Nova

After the resize-confirm command migrates the VM00 to Compute1, the compute0 shutdown off, and VM00 and VM01 on the controller cannot be

The line communicates.

The reasons are as follows:

Before the virtual machine migration, the internal related network of the two compute is composed of the following:

On the compute0 there are router and DHCP server namespace, if the VM00 migrated to VM01, shutdown off Compute0, from the point of view, there is no router namespace,

So the routes to VM00 and VM01 are different, and the associated network composition after the migration is as follows:

Why does DHCP server still exist, this may be related to OpenStack installation, it is possible that DHCP server is HA mode at the time of installation, then you will certainly ask why I know that DHCP server is present

of it? I create a instance, and use DHCP mode, the new instance is assigned to the IP address, and the IP netns is used to see that there is indeed a namespace generation of DHCP.

So how can you shut down the COMPUTE0 and make the network still working? Naturally think of the router also migrated past, the solution is similar, using the HA router, know at the time of creation using the--ha true option, using the

After HA router, the two compute networks are composed of the following:

After the HA router is created, a router namespace is also created on the Compute1, and after the VM00 is migrated, compute0 shutdown, the network composition diagram of the two compute is as follows:

Compute1 's router will be in the compute0 in a few seconds to replace the work to ensure that the network connectivity, if the compute1 on the creation of a VM,DHCP server name namespace will also

Create, stating that the new DHCP server can continue to work.

(Because the company can not send the blog post, unable to post the contents of the operation, can only be recorded in the brain, back to the summary of this blog post)

Related tests for neutron of OpenStack (JUNO)