minikube helm chart #28
Conversation
vivian-rook
force-pushed
the
T301469
branch
3 times, most recently
from
2ae5868
to
963b9fd
Compare
|
I will excuse myself from this code review, because I don't know much about k8s, minikube or helm. |
There was a problem hiding this comment.
I'm not able to set this up on my local as my 8G RAM system gets bogged down on running docker builds within minikube and the kube apiserver crashes. I'll continue to use the docker-compose dev env as it's quite more lightweight.
However, would be great to see if we can get this working on quarry-kube.wmcloud.org. Leaving some comments from that perspective. I see lots of places in the yaml where we'd need additional conditions to differentiate between dev, prod, and the quarry-kube staging setups. Would be nice to inject the values from a file instead of using if conditions.
|
Thank you for the review. A kind of more meta comment. When I first made this I was assuming that the docker-compose env would be replaced by a minikube env. I'm no longer convinced that this is the right immediate path. So I think this is more about having a minikube env to make sure that we would expect the same to work in prod (Or the whole ticket could shift towards prod only, as it is easy enough to deploy multiple clusters in prod now and test there). There are some disadvantages to this method, namely that dev (docker-compose) and prod wouldn't be mostly the same, that has been a thing in the past, where code on docker-compose works, but on the vms does not. Regardless seems like a bridge for another day. |
vivian-rook
force-pushed
the
T301469
branch
3 times, most recently
from
3739057
to
8c713a0
Compare
modification of quarry to allow for deployment to minikube. Additional changes will be needed when k8s is available in production to allow for production to be deployed to k8s. ** Do not deploy this ** Bug: T301469
vivian-rook
force-pushed
the
T301469
branch
3 times, most recently
from
e65a7e3
to
9f539e4
Compare
|
I think with this the current VM setup shouldn't notice any difference, and we have a working minikube deploy, which acts as a good starting point to introduce a magnum deploy. Any other views before we merge? |
modification of quarry to allow for deployment to minikube. Additional changes will be needed when k8s is available in production to allow for production to be deployed to k8s.
** Do not deploy this, not yet ready. **
Bug: T301469