CloudVM instances are not resumed after being paused

[frisso]

A persistent instance has the interesting property of supporting "pause/resume" actions. In a nutshell, the VM can be suspended, hence releasing hardware resources (e.g., CPU; memory), then resumed when needed.
Unfortunately, the "resume" action does not work anymore.

Steps to reproduce:

• start a persistent image (e.g., Cloud Computing: Kubernetes within the "Experimental workspace")
• move into the "Active" section of the dashboard, and click on the "pause" button
• wait a couple of minutes (just to allow everything to be saved correctly on disk)
• within the same section, click on the "Start" button

The instance is never restarted, it hangs forever.

[AleTopp]

Hi @frisso , this would be a big problem, but I didn't notice the issue until now.
I didn't succeed in reproducing the bug. I tried with the VM in the example (also another), and despite taking 2min 17s to start the first time, the VM was running, and I was able to pause it and restart it, both from the Active and from the Workspace tab.

Do you have an instance that is not starting on your namespace that I can inspect?

[frisso]

Dear @AleTopp ,
according to @air-31 , there are two problems (funny, one bug report seems to have produced the identification of two independent bugs):

• First, a problem related to personal workspace.
• Second, it could be because it's a CloudVM type of instance. Ilkhom discovered that, for some reason, when such instances are started, their status (part of their manifest) is not being forged/updated, so the frontend doesn't discover that they're in the Ready state, thus never allowing to connect to them.

In any case, I've currently two instances that are "pending" and do not restart after being paused:

• instance "instance-rfzss", which is a cloud VM belonging to the "Cloud Computing 2025/26" course, template "Ubuntu 22.04 - CloudVM"
• instance "instance-96wp7", again a CloudVM, started in my personal workspace.