Surprising behaviors with ecosystem file

[diwakergupta]

I'm trying to use pmdaemon to run multiple bun apps. Here's what my ecosystem file looks like:

{
  "apps": [
    {
      "name": "app1",
      "cwd": "path/to/app1",
      "script": "bun",
      "args": ["src/index.ts"],
      "port": "4001",
      "env": {
        "NODE_ENV": "production"
      }
    },
    {
      "name": "app2",
      "cwd": "path/to/app2",
      "script": "bun",
      "args": ["index.ts"],
      "port": "4002",
      "env": {
        "NODE_ENV": "production"
      }
    }
  ]
}

I also have a systemd unit file to run pmdaemon automatically on system boot. Looks like this:

[Unit]
Description=PMDaemon Process Manager

[Service]
Type=forking
ExecStart=/path/to/pmdaemon start --config /path/to/ecosystem.json
ExecReload=/path/to//pmdaemon reload all
Restart=always

[Install]
WantedBy=default.target

I'm running into a lot of surprising behavior, both inconsistent with the docs and my general expectations. Here are just a few:

• After initial setup, pmdaemon --config ... start works. I expected that, correspondingly, stop would also stop all services. But that fails with: error: the following required arguments were not provided: <IDENTIFIER>. Expected behavior: if start starts all services, stop should also stop all services.
• If I manually stop each service, and then call start again, that fails with: Failed to start process 'app1': Process already exists: app1. Expected behavior: one should be able to start - stop - start - stop services
• The only way work around I found for the above state was to delete the services (again, one at a time) and call start again. Expectation: delete should have consistent behavior with start when using an ecosystem file, unless a specific identifier is explicitly provided
• This behavior persists across reboots. For instance, say I have a running system with 2 services managed by pmdaemon. The system is rebooted. But pmdaemon fails on reboot with the same process already exists error. I verified that the PIDs listed via status are non-existent. Expectation: pmdaemon restarts services as expected on a reboot

Overall this is a very promising project and I'm excited for a better alternative to pm2, but right now all these weird behaviors make it very unpredictable and difficult to use.

[diwakergupta]

BTW, I'm also noticing that calling stop doesn't actually stop or kill the underlying process.

[entrepeneur4lyf]

Which OS? Platform?

[diwakergupta]

Which OS? Platform?

Ubuntu 24.04, x86_64