RunPod Support Request Critical Boot Loop / Command Override Issue
CRITICAL: Persistent Boot Loop on Custom Template (ID: {2yxqgl3qiz}). Server ignores custom startup command.
Dear RunPod Support Team,
We are experiencing a critical boot failure and crash-loop on our custom ComfyUI deployment in the Community Cloud. The issue is a persistent conflict where the server fails to execute the correct startup script.
1. Environment Details (Latest Attempt):
Template Image (Base): runpod/comfyui:latest-5090 (or similar).
Our Template ID: {2yxqgl3qiz}
Network Volume ID: {lza6mbiiod}
Data Center: EU-RO-1.
Volume Mount Path: /workspace (Set in template).
2. Problem Diagnosis:
The Pod fails within seconds because it is running a corrupted startup script from the ephemeral root filesystem, instead of the clean script on the mounted volume.
Expected Command (Set in Template): bash /workspace/start.sh (The file on the persistent volume).
Actual Script Executed (Observed in Logs): /start.sh
Fatal Error in Log: /start.sh: line 234: uv: command not found
This indicates that the built-in entrypoint of the container image is executing its own internal /start.sh which is corrupted or wrong, and ignoring the Container Start Command defined in our template.
3. Required Action:
We have verified that the correct, clean startup script exists at /workspace/start.sh on our volume.
Please investigate the template execution chain for the image runpod/comfyui:latest-5090 and advise on how to force the container to execute the command defined in the template (bash /workspace/start.sh) without being overridden by the image's internal /start.sh or a shell fallback.
Thank you for your assistance.
Dear RunPod Support Team,
We are experiencing a critical boot failure and crash-loop on our custom ComfyUI deployment in the Community Cloud. The issue is a persistent conflict where the server fails to execute the correct startup script.
1. Environment Details (Latest Attempt):
Template Image (Base): runpod/comfyui:latest-5090 (or similar).
Our Template ID: {2yxqgl3qiz}
Network Volume ID: {lza6mbiiod}
Data Center: EU-RO-1.
Volume Mount Path: /workspace (Set in template).
2. Problem Diagnosis:
The Pod fails within seconds because it is running a corrupted startup script from the ephemeral root filesystem, instead of the clean script on the mounted volume.
Expected Command (Set in Template): bash /workspace/start.sh (The file on the persistent volume).
Actual Script Executed (Observed in Logs): /start.sh
Fatal Error in Log: /start.sh: line 234: uv: command not found
This indicates that the built-in entrypoint of the container image is executing its own internal /start.sh which is corrupted or wrong, and ignoring the Container Start Command defined in our template.
3. Required Action:
We have verified that the correct, clean startup script exists at /workspace/start.sh on our volume.
Please investigate the template execution chain for the image runpod/comfyui:latest-5090 and advise on how to force the container to execute the command defined in the template (bash /workspace/start.sh) without being overridden by the image's internal /start.sh or a shell fallback.
Thank you for your assistance.
