Oleg Sidorov

This task needs a generic example of a workload that could use stateful sets so Shipper can eventually advertise this model.

This proposal is an very interesting topic from Shipper's contract perspective. Say, application A has 2 releases: R1 (deactivated) and R2 (active). In essence, Shipper guarantees to have transitioned application...

> if step 0 and step 1 have totally disjoint cluster selections, does that mean that advancing from 0 to 1 resets the clusters from step 0 to baseline? The...

We made some little progress on this problem and opened a corresponding PR: https://github.com/helm/helm/pull/6010. If this change will appear in helm branch 2, we will be able to bump Shipper...

A quick update on the issue: the fix is not coming any time soon. The change caused a major regression in numeric functions (see https://github.com/helm/helm/issues/6708 for more details). Our attempt...

A new hope to get this fixed comes with this PR: https://github.com/helm/helm/pull/6888. I also created another issue https://github.com/bookingcom/shipper/issues/228 as if the PR gets merged into Helm upstream, we will have...

Hey @harrysingh, after a few attempts to fix the issue the fix was not accepted by Helm folks: in the end it looked reflect-monstrous, I admit. Neither me nor the...

An example of the output I've come up so far: Happy state: ``` reports: - achievedPercent: 100 availableReplicas: 2 cluster: docker-desktop summary: |- 2 container(s) [nginx(x2)] are in status: ContainersReady:True...

Overall, I'm quite positive on moving on with this change. Our preliminary tests indicate the baseline functionality works just fine.

@kanatohodets The current progress of kustomize adoption in Helm seems to be at the very early stage and I'm quite sure it might transform into something completely different-looking in the...