This message was deleted.
# general
s
This message was deleted.
l
It has happened sporadically to me. Sometimes it's because there's two dependencies on the plugin, sometimes it just gets confused. Are you volume/bind mounting your ~/.pulumi directory? If you're using different home directories at different times then you may confuse Pulumi.
And if you are bind-mounting and the .pulumi directory has the wrong owner, that can also cause problems.
h
Im using it as a base image and extending from it within the dockerfile. Your comment about the two dependencies thing has given me something to look into however
l
Are you aware of the language-specific images?
Much smaller
h
Yeah but ive had issues with them in the past but might try it out to see if it fixes this issue.
FYI i fixed this issue, still not 100% sure but its something to do with the stack not being deployed at all yet and running it in the docker container which means that the 'state' (output of
pulumi stack export
doesn't have an aws provider linked to it.) This appears to be required for the docker image to download the provider plugin deploying from my local machine through pulumi fixed the issue in the docker container as the state had been 'created'
b
This is interesting insight, which language were you using?
h
typescript, after discussing internally I have a feeling it may be due to how npm install is being run so going to try another fix today
ok yeah, so the root causes was not running npm install in the main pulumi image inside the dockerfile (since we were using a multistage build) which meant the aws provider wasn't installed and then since it wasnt listed in the "state" it didnt install at runtime and so just failed. tldr; needed to run
npm install
in the pulumi image rather than a multistage build
👍 1