Vũ Quốc Phong

Based on the logs, I guess the cause is: - first build, `RUN echo abc` doesn't cause file system change -> no snapshot was done - next build, cache of...

I think you can use __personal access token__ or __deploy token__ to workaround the issue, since these token will be same build after build. https://docs.gitlab.com/ee/user/packages/npm_registry/#authenticate-with-a-personal-access-token-or-deploy-token

Which environment are you using to run kaniko? If it is EKS, or AWS Code Build, we can consider using other solution to configure AWS role I think.

Also I'm not sure if this repo is still maintained or not, but I hope this PR can help people that faced the same issue as me.

I notice that this issue happens when your endpoint has particular length. For example, both `curl localhost/version` and `curl localhost/versioM` (endpoint length = 7) cause segfault. in my case, `curl...