TLDR

I can't write into my volumes and when I try to put my service to sleep the container never stays alive long enough for me to copy files into it.

The service goes up, prints the sleep, then immediately exits. Every attempt to rsync hits “read-only” / "not running” even though the volume is attached.

What I’m trying to do

Seed a Railway volume from inside the service context (copy /app/analysis/data and /app/analysis/stress_tests/output → /workspace/data/{analysis_data,stress_outputs}), then restore my normal Start Command.

What works

• Build succeeds (Nixpacks detects Python).

• Service shows Active after deploy with a simple hold command.

What’s broken / confusing

1. No runtime logs even when the Start Command prints a lot and sleeps:

   bash -lc '
   set -euo pipefail; set -x
   echo "[seed] start"; mount | grep workspace || true
   mkdir -p /workspace/data/analysis_data /workspace/data/stress_outputs
   rsync -av /app/analysis/data/ /workspace/data/analysis_data/ || cp -R /app/analysis/data/. /workspace/data/analysis_data/
   rsync -av /app/analysis/stress_tests/output/ /workspace/data/stress_outputs/ || cp -R /app/analysis/stress_tests/output/. /workspace/data/stress_outputs/
   du -sh /workspace/data/analysis_data /workspace/data/stress_outputs || true
   echo "[seed] sleep"; sleep 600
   '
   

   • Deploy Logs tab shows “No logs in this time range.”

   • Details shows Deployment successful (Initialization, Build, Deploy all green).

2. SSH refuses to attach even while the service is Active and running a long sleep:

   railway login
   railway link --project creative-unity
   railway ssh --service "Daily-Ladder"
   -> Expected welcome message… "Your application is not running or in a unexpected state"
   

   (I also tried selecting the project/env via prompts; same result.)

3. Running seeding commands locally with railway run --service … clearly executes on my Mac (no /workspace), so I understand that’s not the right context...I’m trying to run inside the container.
   
   What I’ve already tried

   • Different Start Commands: short seeding, verbose seeding with set -x, and pure hold:

     bash -lc 'echo "[hold] container up"; sleep 1800'
     

     Build goes green → service shows Active. Still no runtime logs; SSH still says “not running”.

   • Verified requirements.txt exists at repo root; Nixpacks detects Python; no Dockerfile in use.

   • Re-deployed multiple times (some earlier deploys failed at “create build plan,” later ones succeeded without changes).
     
     What success looks like

     • Ultimately, I am aiming to get scheduled pipelines running cleanly inside Railway. The missing piece is that both services need my repo’s pre-computed parquet files/data available inside their own attached volumes