coder / coder/envbuilder

Full Image Cache instantly completes

Open
#495 0 comments 0 reactions 0 assignees View on GitHub
bug envbuilder needs-investigation
Dominant language
Go
Stars
300
Forks
64
Avg merge
20m
Merged PRs (30d)
1

Description

We are running envbuilder to compile an image from a dockerfile and the config file.

Run without caching everything works as expected its just slow

Run with a cache repo, it creates the image, pushes the image and runs correctly ONCE.

Subsequent launches with the cache enabled will do the usual file extractions and start the image but the layers will not be correct, some of the run commands seemed to not have executed. For example a simple rm in a run command either did not run or the files were recreated by the base layer.

Run with full image caching the image gets a dry run and found in the cache (as long as i did it once with the layer cache)
Unfortunately the image immediately completes after that and the pod gets restarted by the deployment.

When i manually create a pod with the image from the cache repo it works, the envbuilder binary exists and all run commands got executed correctly.

Heres the log, everything looks perfectly fine if only it would execute the entrypoint properly
[envbuilder.log](https://github.com/user-attachments/files/25622697/envbuilder.log)

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with envbuilder.log and reproduce the Dockerfile plus config-file flow using full image caching, comparing the cache-hit launch with a manual pod using the cached image. Trace the envbuilder binary's entrypoint behavior after the full-image cache hit. Done means the cached image runs its entrypoint and all Dockerfile RUN commands instead of immediately completing.

Written by the indexing model from the issue text.

Assessment

Tech stack
docker, go, kubernetes
Domain
build-system, devops, infrastructure
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.