controllers: scale down ocs-client-op csv in non-provider mode #474
Conversation
t0: odf-op w/ X.Y.Z is installed and all dependecies at same version will be installed t1: odf-op is upgraded from X.Y.Z to X.(Y+1).0 and update will happen for odf-op unless there are explicit not upgradeable conditions set t2: dependencies will first get updated from X.Y.Z to X.Y.(Z+1) if the channel has the update, now if one of the dependencies is stuck odf-op keeps trying to check for upgraded install plan which will not be created due to previous install plan for dependencies not being fulfilled. now, odf-op always tries to scale down client-op if not in provider mode in above scenario t1 & t2 aren't strictly linear and odf-op ended up scaling down X.Y.Z version of client-op but not X.Y.(Z+1) which will never reach running state if cluster is configured as non-provider mode. current PR makes sure that client-op csv & deployments are always scaled down for all non-provider cases. Signed-off-by: Leela Venkaiah G <lgangava@ibm.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: leelavg The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
fix for https://bugzilla.redhat.com/show_bug.cgi?id=2311857, I tested wrt expectation and scenarios that I could think of, pls review and surface any left out paths orrefactoring efforts. thanks. |
There was a problem hiding this comment.
Hi Leela, Earlier we decided that we will scale it down via the CSV. What is a need of doing it via the deployment now. With this fix we will be scaling down from both the places. IMO we should scale down there itself instead of having a new code.
Here is a PR https://github.com/red-hat-storage/odf-operator/pull/450/files
|
|
We had a discussion over a call and we decided to close the BZ https://bugzilla.redhat.com/show_bug.cgi?id=2311857#c7 |
t0: odf-op w/ X.Y.Z is installed and all dependecies at same version will be installed
t1: odf-op is upgraded from X.Y.Z to X.(Y+1).0 and update will happen for odf-op unless there are explicit not upgradeable conditions set
t2: dependencies will first get updated from X.Y.Z to X.Y.(Z+1) if the channel has the update, now if one of the dependencies is stuck odf-op keeps trying to check for upgraded install plan which will not be created due to previous install plan for dependencies not being fulfilled.
now, odf-op always tries to scale down client-op if not in provider mode in above scenario t1 & t2 aren't strictly linear and odf-op ended up scaling down X.Y.Z version of client-op but not X.Y.(Z+1) which will never reach running state if cluster is configured as non-provider mode.
current PR makes sure that client-op csv & deployments are always scaled down for all non-provider cases.