Using Terminal in Jupyter

This page focuses on issues when using the Terminal emulator (terminado) when running Jupyter or JupyterLab on notebook.nevis.columbia.edu.

In general, you're better off using a separate terminal program instead of the one in Jupyter. Still, there are times when having a terminal inside a web browser can be handy for quick file manipulations, running UNIX utilities like sed and grep, etc.

No X11

The Jupyter terminal does not include the X11 windowing system (which is difficult to emulate inside a web browser). Therefore, any programs that use X11 (such as ROOT and xterm) will not work.

Even if you immediately use ssh to connect to your workgroup server from notebook, X11 will not be available.

notebook runs CentOS 7

If you want to run programs on notebook (not recommended!), note that notebook uses CentOS 7 as an operating system; most of the Nevis particle-physics systems run Scientific Linux 6. While the differences between CentOS and Scientific Linux are minor, the differences between versions 6 and 7 are significant. It's not likely that a program compiled for one version will run on another.

No condor on notebook

The condor batch management system is not available on notebook. The reasons:

  • To encourage you to immediately ssh to a system with condor.

  • There are three condor batch clusters at Nevis: general, Neutrino, and ATLAS Tier3. The notebook server has no way to offer a choice as to which one should run your job.

  • None of the resources required for your job would be located on notebook; they would be in your home directory or file server. Submitting a job from notebook would mean all I/O requests would pass through notebook (see condor_shadow) slowing all the machines involved unnecessarily.

  • See the above reason: Even if your job ran on notebook, it probably wouldn't run on any of the Scientific Linux 6 machines on the cluster.

notebook is a shared resource

I know it's tempting to look at the memory and processor queues on notebook and think that they're just sitting there for you to use. But notebook is used by all the particle-physics groups at Nevis, and by all the summer students from June through August. If you clog the system with your jobs, you're slowing down everyone, not just those in your working group.

If you're having trouble finding processing resources for your jobs, talk with your research group. Please don't slow everyone else.

Terminal sessions are persistent

Just closing the browser window or tab with terminado running does not exit the session. If you click the Running tab near the top of the Jupyter home window, you can see all the terminal and kernel sessions you're running. These sessions will persist even if you quit your web browser. To turn off a terminal session, do one of the folliowing:

  • Type exit in the terminado window.
  • Click the Shutdown button in the list of processes on the Running page.
  • Click the Control Panel button, then click *Stop My Server" (which will shut down all your Jupyter processes).

Persistent terminal sessions are a nice feature, but you don't need Jupyter for them. The screen and tmux commands are two different ways to create persistent sessions using a standard Terminal program.

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r2 - 2018-07-28 - WilliamSeligman
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2018 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback