Default constraints make no sense on MAAS
Bug #1064291 reported by
Robie Basak
This bug affects 8 people
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Release Notes for Ubuntu |
Invalid
|
Undecided
|
Unassigned | ||
| pyjuju |
Triaged
|
Low
|
Unassigned | ||
Bug Description
I have a MAAS cluster consisting entirely of ARM nodes. I run "juju bootstrap" against this environment, and juju requests amd64 nodes and then fails. This works as expected. But the default does not make sense here.
I understand that on EC2 a reasonable user would expect some sensible instance type instantiated by default, and this is what the default constraints currently achieve.
But with MAAS, I think a reasonable user would expect no constraints to be applied by default.
Does this mean that default constraints need to be per-provider?
Other users affected: those with i386, or with mem < 512M.
Workaround: use: juju bootstrap --constraints "arch=any mem=0"
| tags: | added: arm |
Revision history for this message
| Robie Basak (racb) wrote | #1 |
| summary: |
- Default architecture constraint of amd64 makes no sense on arm + Default constraints makes no sense on MAAS |
| summary: |
- Default constraints makes no sense on MAAS + Default constraints make no sense on MAAS |
| Changed in ubuntu-release-notes: | |
| status: | New → Invalid |
Revision history for this message
| Robie Basak (racb) wrote | #2 |
| description: | updated |
| description: | updated |
| Changed in juju: | |
| importance: | Undecided → Low |
| status: | New → Triaged |
To post a comment you must log in.
This actually applies to other default constraints as well. Why should MAAS+juju require 512M by default? If I set up a MAAS cluster of only nodes with 256M, then it won't work by default as it is. I accept that this is unlikely, but it does demonstrate that default constraints don't make sense on MAAS at all.