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
CLI drops (some?) errors on the floor #1762
Comments
Bizarrely, now I can't even update or destroy this stack, and get 500s from the service each time I try. I'm not sure if it's related to the above failure. Here is the log:
/cc @chrsmith |
Got hit by this again just now:
Inspecting the logs reveals that my Azure token has expired:
|
I've hit this in a handful more cases today. This definitely needs a fix before 0.15 is ready to go.
|
FWIW - I’ve also hit this on every error (I think so far all things originating from the resource provider). |
This test is failing in Travis, and is not debugable due to pulumi/pulumi#1762. Disabling for now. Tracking reenabling this with #130.
There are potentially two different issues here. The first one is that the implementation of I'm not sure what's going on with Joe's original Azure example. Looks like it's probably a similar issue. Either way, a band-aid to make this much better until we tackle #1712 in M17 shouldn't be too bad here. |
It seems weird to me that the azure implementation of |
I have a feeling this isn’t just about Check. I saw this for several failures coming from the provider which I’m fairly certain were errors from Update or Create operations as well. I also hit this for both Azure and Kubernetes this afternoon. |
Here's the state of our error reporting:
The de facto contract that the step generator has with the plan executor is that it has already done all error logging that needs to be done before returning an error. We should strive to centralize our error reporting logic in one place, either in the step generator or in the plan executor. In practice this would mean that the step generator always logs every single error that it sees, or that the plan executor logs every error that comes out of step generation. |
While we do need to focus on the CLI error reporting issue, the reason the service fails the call is because it looks like we are being sent a duplicate metadata key with the update. We'll probably still want to fail the request with a 400, but we can provide a better error message from the service using the expanded ErrorResponse type. |
This test is failing in Travis, and is not debugable due to pulumi/pulumi#1762. Disabling for now. Tracking reenabling this with #130.
I have a resource that is failing its Check, and the CLI simply fails with:
The logs clearly show the failure coming from Check that never gets displayed:
This is on 0.15.0-rc2 bits:
The text was updated successfully, but these errors were encountered: