You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
render --docker: the first image build gets killed after 10 minutes and the error says to check Docker is running #5378
The first time you run render --docker, the CLI builds the renderer image with a 10 minute timeout (timeout: 600_000 on the docker build call in packages/cli/src/commands/render.ts). If the build takes longer, it gets killed and you see spawnSync docker ETIMEDOUT plus "Check Docker is running: docker info". Docker was running fine. The build was just slow.
Minimal reproduction
Any project, no cached hyperframes-renderer image, Docker VM with 2 CPUs (colima's default):
Here one first build went past 10 minutes and got killed during npm install -g hyperframes (step 10/13). Another one, for 0.8.143, took about 8 minutes, so it depends on the machine and the connection.
Expected behavior
The build finishes, or the error says it timed out and that running the command again picks up from the cached layers.
Actual behavior
✗ Docker image build failed
Failed to build Docker image: spawnSync docker ETIMEDOUT
Check Docker is running: docker info
Running the same docker build by hand with no timeout finishes, and after that render --docker works.
Environment
hyperframes 0.8.78 and 0.8.143, macOS 26 on an M4 Pro, colima with 2 CPUs and 4 GiB, Docker 29.4.0.
Additional context
Either dropping the timeout for the build (the output already streams unless --quiet) or catching ETIMEDOUT with a message that says what happened would fix it. Happy to send a PR for whichever you prefer.
I'll take this one. I'll catch the timeout and say what happened, and that running it again continues from the cached layers. Leaving the 10 minute limit as is unless you'd rather raise it.
Activity
diegolopezrm commentedon Oct 10, 2026
I'll take this one. I'll catch the timeout and say what happened, and that running it again continues from the cached layers. Leaving the 10 minute limit as is unless you'd rather raise it.