search

SUSE / catapult

SCF and KubeCF CI implementation
Apache License 2.0
15 stars 9 forks source link

fix cfo upgrade command for CAP 2.1.0 #285

Closed prabalsharma closed 4 years ago

viccuad commented 4 years ago

@satadruroy @prabalsharma Could you expand what gets fixed with this?

The change from --reuse-values to --set "global.singleNamespace.name=scf" should have no effect, as that was already the case, no?

And adding 3 times helm list -A kubectl get pods -A only adds more cruft in my opinion. If we need to wait for something, we would be better served by adding a sleep with its corresponding comment.

satadruroy commented 4 years ago

@viccuad --reuse-values for cfo upgrade fails because there seem to be sufficient template differences between the operator version corresponding to 2.0.1 and the current version. Besides, global.singleNamespace.name=scf is what is used for current operator version whereas what was being used in previous published version was global.operator.watchNamespace=kubecf

The rest of the comments seem valid - we can perhaps take care of them in a separate PR but I'll let @prabalsharma comment.

viccuad commented 4 years ago

@satadruroy thanks! I didn't recall the config differences between 2.0.1 and current. On top of that, it seems that --reuse-values doesn't get checked by the schema of the chart version that was deployed and matches the values, but the schema of the chart you are upgrading, which could be dangerous. And maybe we shouldn't use it if we are upgrading between different versions of the chart (see https://cloudfoundry.slack.com/archives/CQ2U3L6DC/p1601314809489600),

viccuad commented 4 years ago

simplified output of the upgrade target in https://github.com/SUSE/catapult/pull/288.