Jordan Moldow

> this needs to be done recursively: after releasing changes to the vtk, the vtk will schedule changes on its vts children which then need to be released; then each...

Are there any kinds of alters that don't go through the rolling update process? (e.g. increasing the number of replicas?)

#204 makes it so that you can use the `k8s` topo, though it doesn't implement some of the quality-of-life improvements suggested in this thread. In your cluster spec, you have...

@shlomi-noach Would a feature like the following make sense? I'm not sue if you've already considered/implemented this or not, but based on the existence of this PR, I'm _guessing_ it...

> If a migration is `-declarative`, then it is idempotent, whether `-singleton` or not. It's not automated to eventually run -- but if the user submits it again, then it...

> The first, they apply, the second, they realize "nothing to do". 👍 I take it that, even though you said "apply", this would also behave the same way even...

In the case you describe (second migration is requested on the tablet, after the first identical migration has already succeeded on the tablet), we should probably always re-queue the second...

> The issue with UX here is that this is going to be inconsistent. Quick migrations behave differently than long running migrations. If the migration was fast enough to complete...

> I feel like this is exploding/branching too far with if-else scenarios [....] I want to reduce overall complexity. Sure, that makes sense to me, as far as the "handling...

That perhaps makes sense for the images that contain multiple large binaries. But could you still make it such that small client-side binaries remain small, and don't contain all of...