[Bug]:External plugin does not start from docker automatically #1633
Description
🐞 Bug Summary
Applicable to external plugins bootstrapped from cli.
After the makefile dockerizes the plugin and if we try to run the docker image, the plugin seems to start and then exit without errors (i.e., mcp server does not start up). But when if we change the docker entry point, exec into the container and run the run_server.sh script, everything starts up correctly and plugin mcp server starts up.
At first thought it was something to do with .env files or my environment issues, but seems to be happening consistently for newly generated external plugins too.
Has anyone run into this before?
🧩 Affected Component
Select the area of the project impacted:
-
mcpgateway- API -
mcpgateway- UI (admin panel) -
mcpgateway.wrapper- stdio wrapper - Federation or Transports
- CLI, Makefiles, or shell scripts
- Container setup (Docker/Podman/Compose)
- Other (explain below)
Affects external Plugins
🔁 Steps to Reproduce
- Bootstrap external plugin from template
- Run dockerized plugin
- Does not start, container exits
🤔 Expected Behavior
Container should have started the server
📓 Logs / Error Output
Paste any relevant stack traces or logs here.
🧠 Environment Info
You can retrieve most of this from the /version endpoint.
| Key | Value |
|---|---|
| Version or commit | v0.9.0 |
| Runtime | Python 3.12 |
| Platform / OS | macOS, rhel |
| Container | Docker & Podman |
🧩 Additional Context (optional)
Talked with Teryl: Attaching slack thread with fix.
It may be trying to run as stdio rather than streaming http so it’s automatically shutting down.. there is an env variable that specifies it to use streamable http.
It’s called PLUGINS_TRANSPORT set it to “http”
I can confirm that after setting the PLUGINS_TRANSPORT to http, the server starts up correctly. Logging this issue to track updates to doc or fixes for default container bootup process.