-
Notifications
You must be signed in to change notification settings - Fork 97
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix master CI #1772
Comments
trying to fix this: #1772 Obviously this deserves a proper fix. This commit is just to see if it makes any difference. Signed-off-by: Dimitris Karakasilis <[email protected]>
I'm trying a hack here: #1777 A proper fix would allow us to disable certain providers at runtime (not build time) so that we don't need custom builds just to disable a provider. |
The full list of providers is enabled by default. A solution could be to:
The end result is the same but it allows the user to override the default (hardcoded) list with her own, even an empty one. The only drawback is that the default (full) list will be hardcoded in the binary while now it's set in a yaml (though baked in the image with no way to override). Another solution would be to use some environment variable (e.g. |
Hmm, this is more complex than I thought. The yip code doesn't know which stage it's running. If we run all providers when none is set, then we would be doing that for every stage. Currently it only happens for the |
Trying to fix: kairos-io/kairos#1772 Signed-off-by: Dimitris Karakasilis <[email protected]>
While tests are running with the hackish solution (PR), I also created this to try if the hack works: https://github.com/mudler/yip/tree/allow-excluding-cloud-providers |
A kairos-agent bumped to this version of yip ^, when installed with the following config, it skips the set providers:
so unless we can find a better solution and if it proves to fix the CI failures, we can go with this one. |
The (paid) github runners don't run currently. Until we get them back, this story is paused. We are temporarily back to the action-runner-controller based runners. |
I like this one. Allows us to keep kairos and other using yip the same way, while providing flexibility to disable providers via just a simple env var 👍 |
We can keep this feature even if it doesn't fix our CI issues. Maybe other people find it useful. We'll need to document it (TODO). |
Im just wondering why would that issue would only affect those two jobs. Very very weird. |
si it kind of looks like it found the userdata endpoint and even if the ssh keys do not work and cant get them, it creates the userdata sentinel during the cos-setup-network service:
Userdata doesnt seem to be valid:
But the sentinel gets created anyway, which in turn runs the initramfs stages again, which may impact the user/password:
|
Also notice how the hostname is changed due to the userdata. What Im not sure is why is this affecting this 2 jobs specifically.... |
We get 2 different types of errors:
After trying various things for the SSH issue I tend to believe that our new (paid) Github runners run on Azure and we receive datasource from the azure provider:
The reason I believe that is because I see this right before SSH fails:
(here: https://github.com/kairos-io/kairos/actions/runs/5998712300/job/16268535514)
We should find a way to disable at least the azure provider and see if this happens again.
The text was updated successfully, but these errors were encountered: