if thenĮcho "=" echo " You must run 'vncpasswd' once before launching a vnc session" echo "=" echoįi # launch VNC session VNC_DISPLAY = ` $VNCSERVER_BIN -bs 2>&1 | grep desktop | awk -F: '' ` echo "got VNC display : $VNC_DISPLAY " # make sure this is a valid display if thenĮcho "=" echo " ERROR: No vnc display found" echo " Error launching vncserver: $VNCSERVER " echo "=" echo #!/bin/bash # Modified VNC job script, based on: #SBATCH -J vncserver # Job name #SBATCH -p nodes # Queue name #SBATCH -o %j.out # Output File Name #SBATCH -N 1 # Total number of nodes requested (20 cores/node) echo job $JOB_ID execution at: ` date ` NODE_HOSTNAME = ` hostname -s ` echo "running on node $NODE_HOSTNAME " VNCSERVER_BIN = `which vncserver ` echo "using default VNC server $VNCSERVER_BIN " # Check whether a vncpasswd file exists. Here are the steps to achieve this.įirst of all, add the following lines to your ~/.ssh/config: There is no need to change your local VNC viewer settings and your local ssh config this way. You can then use a VNC viewer to connect to 127.0.0.1:53925, which will be forwarded to the :5908. To avoid this, you can forward the random VNC port number to a fixed port number chosen by you at the login node, and forward the latter to a fixed port number, say, 53925, in your local machine. The disadvantage of the previous method is that you have to change your ssh config based on the non-login node number and the VNC port that are assigned randomly each time you run an VNC server in a non-login node. The method is describe well here and here. One can connect a local port, say, 53925, to a VNC port, say, 5908, at a non-login node through the login node. The connection has to be established through the login node using SSH port forwarding. You cannot log into a non-login node directly, your VNC viewer faces the same problem, it cannot get connect to the VNC server running at a non-login node. The correct way to run a VNC server in a Linux cluster is to run it in a non-login node.
The job of the login node is not to run jobs, including running VNC servers, but to distribute jobs to other nodes.
The easiest way to run an VNC server in a Linux cluster is to run it at the login node. Slurm is a popular job scheduler that helps to distribute jobs to less busy nodes in a Linux cluster. While you cannot log into other nodes directly, you can submit jobs (programs) from the login node to other nodes through a program called job scheduler. There is normally a login node, which you can ssh into. A Linux cluster is a group of Linux machines working as a whole.
0 kommentar(er)
