7 months ago
- Project: steadfast-education
- Deployment frozen at
01:59:48 - Every
COPYstep is cached despite code changes - Need build cache cleared for this service
4 Replies
Status changed to Awaiting Railway Response Railway • 7 months ago
7 months ago
You can disable build layer caching by setting the NO_CACHE=1 service variable, then triggering a redeploy. If the current deployment is stuck, you can cancel it from the deployment's three-dot menu, then use the Command Palette (CMD/Ctrl + K > "Deploy Latest Commit") to start a fresh build.
Status changed to Awaiting User Response Railway • 7 months ago
7 months ago
"The NO_CACHE=1 variable did not fix it. Build still shows ✓ 1 cached on every step including npm run build (built in 1.04s). This is the same cached layer from weeks ago. Need server-side cache cleared for project steadfast-education, service Clear2Close."
Status changed to Awaiting Railway Response Railway • 7 months ago
Status changed to Open brody • 7 months ago
a month ago
Hi! I looked into your Clear2Close deployment issue.
Since "NO_CACHE=1" didn't resolve it, I wouldn't assume the build cache is the root cause yet.
Here's a diagnostic sequence I'd recommend:
- Compare the latest GitHub commit SHA with the commit actually deployed by Railway. Verify the connected branch, root directory, and watch paths.
- In a safe test environment, use Deploy Latest Commit rather than repeatedly redeploying an existing deployment.
- Add a unique revision marker to your source code and expose it through a temporary static file or diagnostic endpoint. Check whether the marker reaches the live application.
- If the marker is missing, inspect your Dockerfile, ".dockerignore", build context, and "COPY" instructions. Also verify that "NO_CACHE=1" was applied to the correct service and environment.
- If the latest commit is confirmed, caching is disabled, and the correct build still fails to deploy, provide Railway Support with the deployment ID, build logs, and the revision-marker test results.
The important distinction is that "CACHED" in a build log does not necessarily mean stale code. We need to establish exactly where the expected revision disappears.
It will depend of:
- Your builder type (Dockerfile, Railpack, or Nixpacks)
-
- The expected Git commit SHA and the SHA Railway deployed
-
- Your Dockerfile or build configuration, if applicable
-
- The relevant build log section
-
- Whether a unique revision marker appears in the deployed app
With those details, we can narrow this down to source selection, build caching, runtime deployment, or application-level caching without deleting the service or risking persistent data.
a month ago
We’re interested in investigating the persistent COPY-cache behavior. If this is still open, could you share the relevant Dockerfile or public repository so we can verify the cause and propose a safe cache-busting fix?
