Skip to content

Conversation

jackhodgkiss
Copy link
Contributor

@jackhodgkiss jackhodgkiss commented Nov 24, 2022

  • Fix _bootstrap_user for all groups to use cloud-user
  • Use pulp proxy to avoid slow downloads
  • Include seed within growroot_group as seed uses LVM images
  • Enable Octavia backed by OVN
  • Bump cephadm
  • Reduce MTU on external_network to enable communication with VMs on admin-vxlan

@jackhodgkiss jackhodgkiss self-assigned this Nov 24, 2022
This reduction is required as VMs placed on the `admin-vxlan` network
cannot be accessed.
@jackhodgkiss jackhodgkiss marked this pull request as ready for review November 24, 2022 11:42
@jackhodgkiss jackhodgkiss requested a review from a team as a code owner November 24, 2022 11:42
jackhodgkiss added a commit that referenced this pull request Nov 24, 2022
jackhodgkiss added a commit that referenced this pull request Nov 24, 2022
jackhodgkiss added a commit that referenced this pull request Nov 25, 2022
jackhodgkiss added a commit that referenced this pull request Nov 25, 2022
jackhodgkiss added a commit that referenced this pull request Nov 25, 2022
jackhodgkiss added a commit that referenced this pull request Nov 25, 2022
@jackhodgkiss jackhodgkiss changed the title Multinode load balancers Multinode Wallaby Nov 25, 2022
# External network
external_cidr: 192.168.38.0/24
external_mtu: 1400
external_mtu: 1350
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think @sd109 hit this with his aufn work, and possibly @darmach with his magnum work. It's because we're using VXLAN networks with OVN, but internally it sometimes uses GENEVE tunnels which have a larger header size.

One option is to switch to geneve tenant networks when using OVN, which generally work better.

Comment on lines +18 to +21
# Octavia load balancer configuration
octavia_auto_configure: "no"
octavia_provider_drivers: "ovn:OVN provider"
octavia_provider_agents: "ovn"
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We probably only want this when kolla_enable_ovn is true.

kolla_enable_cinder: true
kolla_enable_cinder_backup: true
kolla_enable_neutron_provider_networks: true
kolla_enable_ovn: true
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Unsure whether we should change this default. Ideally it should be possible & easy to use either OVS or OVN.

kolla_enable_cinder_backup: true
kolla_enable_neutron_provider_networks: true
kolla_enable_ovn: true
kolla_enable_octavia: true
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Perhaps only enable octavia when OVN is enabled, since we don't have the config for the amphora provider in place currently.

markgoddard pushed a commit that referenced this pull request Jan 31, 2023
* feat:  copy `ci-multinode` from PR `#246`

* feat: copy `ansible` from PR `#246`

* revert: configure-aio-resources.yml to `da05599`

* feat: actually copy contents of `ci-multinode`

For some reason the previous commit does not contain the important
changes to the configuration. This commit fixes that.

* feat: include `.automation.conf` from PR `#246`

* revert: `stackhpc-ci` back to `stackhpc/xena`

* feat: update address of `pulp-server`

* feat: bump `stackhpc.cephadm` version

* feat: `haproxy` group has been replaced with `loadbalancer`

* fix: ensure `cephadm_host_labels` is defined

* feat: reduce concurrency and remove load list

* feat: increase `max_microversion`

* fix: restore order to `requirements.yml`

* fix: README has been moved

* fix: remove unused collection

* fix: add missing new line

* feat: disable some unwanted service within ci-multinode

* fix: remove extra blank lines

* feat: revert `TEMPEST_CONCURRENCY` to two

* feat: load `ci-multinode` tempest settings conditionally

Firstly, the skip-list has been renamed from `tempest-full` to `ci-multinode` as this
list would be environment specific. Secondly the `config.sh` script now contains a
conditional for the `ci-multinode` environment allowing for the test list and
concurrency settings to be loaded without impacting other environments.

* fix: use `HTTP` dashboard URL over `HTTPS`

* fix: reduce multinode tempest concurrency

* feat: bump `stackhpc.cephadm` collection version

* feat: remove unecessary comment
Copy link
Contributor

@markgoddard markgoddard left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you need to reduce TEMPEST_CONCURRENCY as you did in xena?

Copy link
Contributor

@markgoddard markgoddard left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

 [dashboard]
-dashboard_url = https://192.168.39.2
+dashboard_url = http://192.168.39.2

@markgoddard
Copy link
Contributor

You'll probably also want to merge in stackhpc/wallaby, maybe in a separate PR.

@jackhodgkiss jackhodgkiss merged commit 5924262 into multinode-env Feb 6, 2023
@jackhodgkiss jackhodgkiss deleted the multinode-load-balancers branch February 6, 2023 12:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants