cpu-cores constraint no longer respected

[Jeff Pihach]

Juju version: 2.3.4-xenial-amd64

When making a request over the API for a 4 core CPU if the call uses the legacy key 'cpu-cores' it will disregard that constraint when requesting a new machine. If you send the request using the 'cores' constraint then it will create a machine with the appropriate number of cores.

This has caused a regression in the GUI: https://github.com/juju/juju-gui/issues/3560 as we must support multiple versions of Juju and there doesn't appear to have been a facade bump for this change.

DOESN'T WORK (cpu-cores)
{"type":"Client","request":"AddMachines","params":{"params":[{"jobs":["JobHostUnits"],"series":null,"parent-id":null,"container-type":null,"constraints":{"cpu-cores":4}}]},"version":1,"request-id":17}

WORKS (cores)
{"type":"Client","request":"AddMachines","params":{"params":[{"jobs":["JobHostUnits"],"series":null,"parent-id":null,"container-type":null,"constraints":{"cores":4}}]},"version":1,"request-id":7}

[Jeff Pihach (hatch)]

The GUI issue is for vsphere but the issue also happens on GCE (possibly others as well)

[Nicholas Skaggs (nskaggs)]

Can you provide a log confirming this? It seems like this is an API issue with your call, but not an issue on the cli.

[Jeff Pihach (hatch)]

@nskaggs the calls are outlined in the original description.

It's also interesting to note that the machine values returned by the megawatcher use 'cpu-cores' and not the newer 'cores'. So it seems like this change was made but never finalized throughout the codebase with an incremented facade.

[Canonical Juju QA Bot (juju-qa-bot)]

This bug has not been updated in 2 years, so we're marking it Low importance. If you believe this is incorrect, please update the importance.