I have a stuck deployment that was kicked off from CI GitHub Pulumi Community #pulumi-cloud

I have a stuck deployment that was kicked off from...

little-cartoon-10569

05/15/2022, 9:42 PM

I have a stuck deployment that was kicked off from CI (GitHub action). It has failed because of invalid resource combinations in the target AWS environment, but now it's just sitting there, waiting for a

pulumi cancel

or similar. Since I don't have command-line access to this stack (it's all locked down via OIDC, only that particular GitHub branch can control this stack), what's the suggested way of cancelling this deployment?

little-cartoon-10569

05/15/2022, 9:42 PM

I'd really like a "Cancel" button in the deployment's pane in the "Activity" tab, as that would just work.

little-cartoon-10569

05/15/2022, 9:43 PM

Failing that, I could set up a special PR or gitops thing to run

pulumi cancel

.. is there an example of that anywhere?

little-cartoon-10569

05/15/2022, 9:45 PM

And the last option would be to set up a back door in the OIDC connectivity.. another role that allows access from dev machines to run

pulumi cancel

. This would be the most general purpose solution, but would also be our last choice. If we're able to have a robust deployment strategy that forbids all unplanned / manual access, that's what we'd prefer.

little-cartoon-10569

05/15/2022, 9:47 PM

A timeout option on a

pulumi up

would also work for us. Since we know how long a full deployment takes, we can use that information at deployment time to get Pulumi to auto-cancel when we go significantly longer. Is that a feature?

brainy-church-78120

05/15/2022, 9:49 PM

docs on custom timeouts: https://www.pulumi.com/docs/intro/concepts/resources/options/customtimeouts/

little-cartoon-10569

05/15/2022, 9:53 PM

Thanks. That's a per-resource timeout. We can configure that on the offending resource this time, since we now know that that's a potential problem. But it won't be a general fix, unfortunately. Is there an equivalent for an entire