instance launch

asked 2019-05-20 02:08:01 -0600

lelunicu gravatar image

hi, just to clarify.

i like to understand what services are involved when an image is created-it means what service will create the vnic vcpu so on inside the image.

when we launch an image-heat read the specification inside yml file and create objects like image-then connect volume to image then create vnic vcpu so on.right?

so neutron will create the vnic inside image? or when launch an instance with volume from yml file behind will be a command like openstack server create --flavor FLAVOR --block-device \ source=SOURCE,id=ID,dest=DEST,size=SIZE,shutdown=PRESERVE,bootindex=INDEX \ NAME

edit retag flag offensive close merge delete

1 answer

Sort by » oldest newest most voted

answered 2019-05-20 03:30:06 -0600

updated 2019-05-20 03:32:36 -0600

VNICs and VCPUs inside the instance are created by the hypervisor that is configured on the compute node where the instance runs, for example KVM. There are no VNICs or VCPUs inside the image. To connect the instance to the network, Nova works with Neutron, to connect it to a volume, it works with Cinder. To get access to an image, Nova gets help from Glance.

Heat creates all objects that you specify in the YAML template by issuing API requests to the corresponding services. For example, to create an object of type OS::NOVA::SERVER, it issues an API request to Nova.

Volumes are not connected to images. Rather, if you launch an instance in the way you specify in the question, image content is copied to a new volume, then the instance is booted from that volume.

edit flag offensive delete link more


so the hypervisor create the vnic inside instance and then the nutron will connect this vnic to a specific network through a port.right? heat will read the content of yaml and will decide on which compute the instance will be lauched.right? then those instances are monitored and in log will be

lelunicu gravatar imagelelunicu ( 2019-05-20 05:08:20 -0600 )edit

written things like when instance was created of the instance failed to create.

lelunicu gravatar imagelelunicu ( 2019-05-20 05:09:40 -0600 )edit

Almost correct. A port is the Neutron abstraction of a network onnection. How the instance is connected depends mostly on Neutron’s core plugin; in case of the ML2 plugin, it then depends on the mechanism driver. For example, via a Linuxbridge plugged into an Openvswitch bridge.

Bernd Bausch gravatar imageBernd Bausch ( 2019-05-20 06:39:21 -0600 )edit

Heat doesn’t decide on which compute node the instance will run. This decision is made by the Nova scheduler. I am not sure to which extent it monitors instances. You can also use the commands openstack stack event list and resource list to see which instance launches were successful.

Bernd Bausch gravatar imageBernd Bausch ( 2019-05-20 06:42:34 -0600 )edit

last thing. if the instance crash or fail then i must look into the logs on that compute server not in controller.right?

heat will use api call to nova to create the instance-then nova will talk to neutron to cinder to glance to attach the network the volume and image to the instance.right?

lelunicu gravatar imagelelunicu ( 2019-05-20 07:59:46 -0600 )edit

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2019-05-20 02:08:01 -0600

Seen: 82 times

Last updated: May 20 '19