sudo snap refresh conjure-up --edge?
hello,
why is this creating 11 machines if i use only 6 ?
thanks
I know one additional machine is for the juju controller (when you bootstrapped) , but we can't reproduce what you are seeing. Can you attach tarball of ~/.cache/conjure-up/?
@battlemidget The controller machine would be in a different model, so wouldn't be counted in the screenshot. However, it does look like the number of units for both etcd and kubernetes-worker were reduced, and conjure-up did not reflect that in the final machine count. There is some refactoring that needs to be done in configapps/gui.py
the. nr of machines was reduced 1 etcd instead of 3 and 2 workers instead of 3.
juju controller its on another machine
can i remove those containers whose ips are not used by units?
Yea if no units are attached you can remove-machine
On the localhost provider, it seems this also happens due to the machines being pre-created without any constraints but when the applications are deployed the placement directives are stripped but constraints left intact, which causes any applications with constraints specified (e.g., kube-worker) to create new machines rather than re-using a pre-created machine.
This stripping of placement seems to be there to prevent issues with things like sub-machine placement, but it would be worth testing if it's even needed any more. If it is, though, (or just as an easy fix) we should just skip pre-machine creation entirely for localhost.
@battlemidget @johnsca It looks like we just had an occurrence of this reported in a CDK issue. Can you confirm?
https://github.com/juju-solutions/bundle-canonical-kubernetes/issues/433
@Cynerva yea this is same, we are reworking this code this cycle.
This has now been addressed