[RFE] support trunk's subport range

Bug #1698500 reported by Yan Songming
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
neutron
Won't Fix
Wishlist
Unassigned

Bug Description

When we use one trunk port,the following step should be done.
1 openstack port create --network project-net-A trunk
2 openstack port create --network trunked-net subport1
3 openstack network trunk create \
--parent-port 73fb9d54-43a7-4bb1-a8dc-569e0e0a0a38 \
--subport port=91f9dde8-80a4-4506-b5da-c287feb8f5d8, \
segmentation-type=vlan,segmentation-id=100

But when we need to use a trunk port for vlan 1000~2000. That need tremendous work to do.
Suggesting add API for subport range.

Tags: rfe trunk
Changed in neutron:
assignee: nobody → Yan Songming (songmingyan)
description: updated
summary: - trunk port range
+ [RFE] support trunk's subport range
Revision history for this message
Kevin Benton (kevinbenton) wrote :

How would you expect this to work with a range? How would it know which sub-ports to use?

tags: added: rfe
Changed in neutron:
importance: Undecided → Wishlist
status: New → Triaged
Revision history for this message
Yan Songming (songmingyan) wrote :

We hope to add a api like this:
openstack network trunk create \
--parent-port 73fb9d54-43a7-4bb1-a8dc-569e0e0a0a38 \
--subport port=auto, \
segmentation-type=vlan,segmentation-id=100,200

That means we we will create 100 net,it's net's type is vlan ,segmentid is 100 to 200.
Each net will create one port to use as a subport, subport's segmentation-type=vlan , segmentation-id=100~200

tags: added: trunk
Revision history for this message
Armando Migliaccio (armando-migliaccio) wrote :

This was proposed and rejected at the time of the spec discussion, as it's quite a deal of orchestration forced on neutron server, for something that is usually and typically scripted. There's greater implication on how to coordinate the creation of ports, handling failure modes etc and for this reason I am soft reject. That said, I'd be open to other people's opinions.

Revision history for this message
Armando Migliaccio (armando-migliaccio) wrote :

BTW: a similar request [1] that did not yield success. If we go down the path of approving both the server side logic can really bloat.

[1] https://review.openstack.org/#/c/421880/

Revision history for this message
Armando Migliaccio (armando-migliaccio) wrote :

Furthermore: your proposal of introducing a CLI call like the following:

openstack network trunk create \
--parent-port 73fb9d54-43a7-4bb1-a8dc-569e0e0a0a38 \
--subport port=auto, \
segmentation-type=vlan,segmentation-id=100,200

is unclear which networks subports must plug into.

Changed in neutron:
status: Triaged → Incomplete
Revision history for this message
Armando Migliaccio (armando-migliaccio) wrote :

Marking incomplete until use case is clarified.

Changed in neutron:
assignee: Yan Songming (songmingyan) → nobody
Revision history for this message
Yan Songming (songmingyan) wrote :

Yes, we can do this by using script. But i think this will increase the API calls,which is bad for performance.
As for network:
"openstack network trunk create \
--parent-port 73fb9d54-43a7-4bb1-a8dc-569e0e0a0a38 \
--subport port=auto, \
segmentation-type=vlan,segmentation-id=100,200 "
This means neutron will create 100 networks, the segmentation is vlan,the segment id is 100~200 automatically.
Each net will create a port for subport.

Revision history for this message
Armando Migliaccio (armando-migliaccio) wrote :

Having a single command deal with the amount of orchestration you're proposing is not good either, as it may lead to inconsistencies etc.

Our design philosophy is to provide users with smaller building blocks that can be assembled to achieve more complex use cases.

Changed in neutron:
status: Incomplete → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.