nova-compute network-device-mtu setting doesn't work for running VMs
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OpenStack Nova Compute Charm |
Triaged
|
Wishlist
|
Unassigned | ||
| nova-compute (Juju Charms Collection) |
Invalid
|
Wishlist
|
Unassigned | ||
Bug Description
I have a requirement to enable jumbo frames on OpenStack, so I'm working through the requirements - one of which is to get the tap/qbr/qvo/qvb devices for a VM to have a higher MTU. After doing a juju set nova-compute network-
However, if I have a running VM, the network devices stay at the old MTU value. I've tried rebooting the old VMs, stopping and then starting, and nothing sets them correctly. The workaround I came up with was to manually set the MTU on the instances interfaces on the nova-compute node.
Testing was done using MaaS 1.8, Juju 1.22.6, Ubuntu 14.04 with latest updates deploying Icehouse and nova-compute charm revno 136. Please let me know if you need any further information.
| tags: | added: canonical-bootstack |
| James Page (james-page) wrote | #1 |
| Changed in nova-compute (Juju Charms Collection): | |
| importance: | Undecided → Wishlist |
| status: | New → Triaged |
| milestone: | none → 15.10 |
| tags: | added: openstack |
| Changed in nova-compute (Juju Charms Collection): | |
| milestone: | 15.10 → 16.04 |
| Changed in nova-compute (Juju Charms Collection): | |
| milestone: | 16.04 → 16.07 |
| Changed in nova-compute (Juju Charms Collection): | |
| milestone: | 16.07 → 16.10 |
| Changed in charm-nova-compute: | |
| importance: | Undecided → Wishlist |
| status: | New → Triaged |
| Changed in nova-compute (Juju Charms Collection): | |
| status: | Triaged → Invalid |
Suggest we tie this into a action to allow the networking on a compute node to be completely reset.