Important information on Remote debugging
Dear users,
As we are seeing many users getting a “Connection refused” error when trying to perform a Remote debugging on aimagelab-srv, here is an important tip to avoid it.
Given that we now have two development nodes (ailb-login-02 and ailb-login-03) it is important to tell SLURM which node you expect your debug session to live on, otherwise SLURM might give you a job on different node from the one you are connected to. You can specify the node you want your debug job to be via the “-w” option in srun. Therefore, you should adjust the bash script that creates the SLURM job for Remote debugging to contain the following line:
srun -Q --immediate=10 -w=<host> --partition=all_serial --account=<account> --gres=gpu:1 --time 60:00 --pty bash
where host is the frontend node to which you are connected (ailb-login-02 or ailb-login-03).
You can read more about this in the updated documentation at https://ailb-web.ing.unimore.it/coldfront/documentation/Zf44P.
If you encounter any issue, do not hesitate to open a ticket at the Help Desk at aimagelab-srv-support@unimore.it
Published: September 4, 2024