UX: Enable cancelling of tasks that are going to fail after a long timeout

[robi]

Now it's something like 120seconds, which is an eternity to wait. if we can bring it down to 10s or so that would be great.

After failure the only thing we can do is retry the tasks, but not cancel, which seems like a miss.

It was nice to have the checkbox for the second task that was failing, which we could uncheck and get things moving again.

[nebulon]

The long wait was intentional here. Users would cancel tasks because they take too long, like a backup or domain changes which due to long TTLs simply do take long. The result was that users end up with broken apps due to cancelling tasks prematurely.

I wonder in your case if there are better fixes. The question might be to see why you want to cancel a task in the first place?

[d19dotca]

@nebulon One use-case I can think of is when a backup is going on and therefore nothing else can really be done (i.e. can't change memory on an app as it has to wait for the other tasks to complete), but then maybe during that time I change my mind and want to increase the memory even more than I originally did, I could cancel the change and make the correct one so it doesn't need to be restarted twice. It's really a minor issue in my mind but would be nice to cancel "waiting/pending" tasks in particular in my use-cases.

[robi]

I wonder in your case if there are better fixes. The question might be to see why you want to cancel a task in the first place?

Task queue management might be one of those better ways.

There will always be some who will shoot themselves in the foot, but a hammer is a tool and your responsibility in how to "apply" it.

That way as @d19dotca mentioned, fixing the order of operations and taking out an unneeded one is much more responsive.

I'd also like to queue up a few operations and have them keep working in the background, while I do more important things than wait for one or more to finish.

[Lonk]

As a developer, I had to uninstall, re-install a lot, and waiting forever for that little "x" to appear to cancel the starting phase, the healthcheck, etc (depends on which phase of development I was on) too forever in "I have an idea to fix this time."

Maybe hardcode Contributors with getting access to that cancellation everytime, orrrrrrr just throw it in with the buttons in the logs. Because we also usually can tell from there. I will say the first time my app worked I didn't know and I left the logs to go uninstall it for my "next test" when I saw it was "Running".

[nebulon]

As a developer you could also use the CLI tool, that has a cloudron cancel command to do it immediately.

[girish]

@Lonk uninstall is always available, never disabled. Unlike other actions.

[Lonk]

@nebulon As a developer, I'm an idiot. I'll be doing that from now on; thanks!

[Lonk]

@girish There's was a lot of times I didn't want to uninstall to re-install (just re-start with some box changes), but when "restarting" them I had to wait for them to finally "Not Respond" at certain points to restart my test of it connecting apps. Obviously because I was working in box out of necessity, my specific situation was a little weird.