Kaniko: standard_init_linux.go:211: exec user process caused "no such file or directory"

[bitsofinfo]

Running gcr.io/kaniko-project/executor:debug 17.1

I have a dockerfile that when i build locally runs fine, however when kaniko builds it, running it yields

standard_init_linux.go:211: exec user process caused "no such file or directory"

what would cause this?

1024 related?

[bitsofinfo]

only reverting to 0.15.0 lets me actually get working images

[donnyv12]

Having this issue as well using the kaniko builder in Google' Cloud Build. Would love to get some eyes on this as it has been a massive pain to debug.

[MarcoSchaar]

facing same issue with version 17.1 but only when building images based on debian
version 18 fixed the issue in my case
thanks

[tejal29]

@donnyv12 are you using the cache=true flag? If yes, please clear the artifact cache.

[tejal29]

@bitsofinfo and @donnyv12 can you please provide some more context to help us debug your issue
e.g.

• dockerfile?
• cloudbuild.yaml

Thanks
Tejal

[tejal29]

ping!

[donnyv12]

Hi @tejal29 sorry for the delay. I am using the cache=true flag and will try it without that.

The docker file is quite large, and is based on a centos image. I'll try clearing the cache and see if that solves it.

[tstromberg]

@donnyv12 - if you get a chance to provide a Dockerfile, I can probably help get to the bottom of this one.

My gut instinct: the programmed called via "RUN" requires libc, which Kaniko lacks. Second best guess is a PATH or architecture difference.

[tejal29]

@bitsofinfo Can you please try this with latest v0.20.0 release and let us know if it still exists.

[beeman]

It looks like I'm running into a similar issue, and I have a repro that I can share:

This tree of my repo with this Dockerfile:

# Use nxgo/cli as the base image to do the build
FROM nxgo/cli as builder

# Create app directory
WORKDIR /workspace

# Copy package.json and the lock file
COPY package.json yarn.lock /workspace/

# Install app dependencies
RUN yarn

# Copy source files
COPY . .

# Build apps
RUN yarn build api

# This is the stage where the final production image is built
FROM gcr.io/distroless/base

# Copy over artifacts from builder image
COPY --from=builder /workspace/dist/apps/api /app

# Set environment variables
ENV PORT=3000
ENV HOST=0.0.0.0

# Expose default port
EXPOSE 3000

# Start server
CMD [ "/app" ]

I'm using the following commands:

make build
make run

[dhstack]

I'm running into this issue too when using the Google Cloud Builders Community Firebase build step.

I don't have a Dockerfile, only a cloudbuild.yaml file that looks like this:

steps:
# Deploy
- name: 'gcr.io/<project-id>/firebase'
  args: ['deploy', '--project=<project-id>', '--only=hosting']

Where <project-id> is the id of my Google Cloud project.

I'm using this command locally with the latest version of gcloud on Windows: gcloud builds submit .

Happy to provide any additional information as needed.

EDIT: For what it's worth, I'm following the official Google tutorial for deploying to Firebase.

[dhstack]

Well I figured it out for me! Since I'm on Windows, I needed to change the line endings of firebase.bash to LF before submitting the community build to my project. More info here and here.

I created an issue for Firebase here: https://github.com/GoogleCloudPlatform/cloud-builders-community/issues/443