-
Spike
-
Resolution: Unresolved
-
Minor
-
None
-
4.16
-
Proactive Architecture
-
False
-
None
-
False
-
OCPSTRAT-1665 - CAPI-based Installer technical debt
-
-
Description of problem:
Between openshift-install create manifests and openshift-install create cluster, users can edit manifests. When editing manifests related to each other, notably the control-plane-machine-set and the master machines (either openshift/99_...master-machines...yaml or cluster-api/machines/10_inframachine...yaml), if the resulting set of manifests has differing providerSpec sections, unexpected/weird things can happen during/after cluster install. See below for more.
Version-Release number of selected component (if applicable):
All, really
How reproducible:
100%
Steps to Reproduce:
1. openshift-install create manifests with a minimal AWS install-config 2. Edit the providerSpec for the master-machines manifests but not the CPMS (or vice versa) 3. opesnhift-install create cluster
Actual results:
Master machines are created by installer according to the master-machines providerSpec; then after install completes they are replaced according to the CMPS.
Expected results:
I mean, yeah, that's the expected result. But it's almost certainly not what the user wanted. Fail `create cluster` with an appropriate message.
Additional info:
Context in super long thread around here.