enable-ha on lxd cluster doesn't evenly spread controllers
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Triaged
|
Medium
|
Unassigned |
Bug Description
juju 2.9.28
After substantiating an LXD cluster with 3 nodes, running `juju enable-ha` after bootstrap didn't evenly distribute the next 2 controllers
```
$ lxc cluster list
+------
| NAME | URL | ROLES | ARCHITECTURE | FAILURE DOMAIN | DESCRIPTION | STATE | MESSAGE |
+------
| repsold | https:/
| | | database | | | | | |
+------
| tebbutt | https:/
+------
| townley | https:/
+------
```
```
$ lxc list
+------
| NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS | LOCATION |
+------
| juju-609cc2-0 | RUNNING | 172.16.8.195 (eth0) | | CONTAINER | 0 | repsold |
+------
| juju-609cc2-1 | RUNNING | 172.16.8.198 (eth0) | | CONTAINER | 0 | townley |
+------
| juju-609cc2-2 | RUNNING | 172.16.8.197 (eth0) | | CONTAINER | 0 | townley |
+------
```
```
$ juju status -m controller
Model Controller Cloud/Region Version SLA Timestamp
controller local localhost/localhost 2.9.28 unsupported 19:18:15Z
Machine State DNS Inst id Series AZ Message
0 started 172.16.8.195 juju-609cc2-0 focal Running
1 started 172.16.8.198 juju-609cc2-1 focal Running
2 started 172.16.8.197 juju-609cc2-2 focal Running
```
Expected behavior would be that the controllers would at least spread to separate cluster nodes, mimicking zone type behavior.
tags: | added: lxd-cloud |
Changed in juju: | |
importance: | Undecided → Medium |
milestone: | none → 3.1-beta1 |
status: | New → Triaged |
Changed in juju: | |
milestone: | 3.1-beta1 → 3.1-rc1 |
Changed in juju: | |
milestone: | 3.1-rc1 → 3.1-rc2 |
Changed in juju: | |
milestone: | 3.1-rc2 → 3.1-rc3 |