I use Colab pro, open a session in a browser and type commands in the terminal. I especially install new software. But when I close the browser, my colab setting are restarted and I have to reinstall all those software again. Is there any way I can keep the software that are installed through terminal?
As you noted, Colab automatically destroys VMs after detecting user inactivity.
Colab Pro+ has a feature called background execution, which is exactly what you asked for: VMs persist after you close your browser. Note that Colab Pro+ costs 5x more than Colab Pro (as of 2022-01-09).
Alternatively, if the process of setting up the environment does not take a long time, I would put all the installation commands in the first cell, using shell access (!apt install my-things) or bash magic (%%bash). Thus, installing the software is done with one cell execution.
Related
My problem is the following: I want to run a Jupyter notebook on my remote desktop and access it via my laptop elsewhere. I have accomplished this, but I can't use my GPU for tensorflow because the GPU-supported version is only installed in my custom, non-base environment. Even though all of my installed jupyter kernels are available, it seems things don't work right unless I run 'jupyter notebook' from within the correct activated conda environment (says "no GPU" even though I select as the kernel the one where tensorflow-gpu is installed).
Is there a simple way of running jupyter notebook from within that environment by a batch script? I also need it to run the notebook on a secondary drive.
I could of course just start up the server while at home and then access it using the token, but that's a little clumsy.
I've found a solution. On windows, in %AppData%\Roaming\Microsoft\Windows\Start Menu\Programs\Anaconda3, there are shortcuts for various Anaconda-related programs, including Jupyter notebook for each environment.
The shortcut for Jupyter notebook for my given env is
`E:\Software\Anaconda3\python.exe E:\Software\Anaconda3\cwp.py E:\Software\Anaconda3\envs\tf E:\Software\Anaconda3\envs\tf\python.exe E:\Software\Anaconda3\envs\tf\Scripts\jupyter-notebook-script.py "%USERPROFILE%".
I modified this to end in '"E:" --no-browser' instead of the userprofile bit and made that into a script. Now when I SSH into the computer and run this script, the notebook is within the correct environment and I have access to my GPU, all on the correct drive, E.
I use remote shh interpreter in PyCharm regularly, using the configured deployment. I often run remote programs from PyCharm GUI (using F5 key), that takes hours to complete (e.g. training a deep net). This unfortunately means that any network outage causes running script to exit and I have to run the script over again. Is there a way to detach the running script so it keeps running? In the sense similar to what screen or nohup is doing? I know I can run it in screen manually via ssh, but it is a bit inconvenient.
Ok. I found out that this feature is not implemented yet. There is however a suggested workaround
I understand the GUI (such as those powered by tkinter) does not work on Google Colab, any alternatives at this point?
Error message
TclError: no display name and no $DISPLAY environment variable in google's colab
To use these notebooks you need to install binary MoebInv libraries and their dependencies.
In short, you simply need to execute it in CoLab or your Ubuntu-18.04 desktop the next cell
So I have Ubuntu 14.04 Installed on my system with a few "tweaks", mainly ccsm. I also have google chromium browser installed as well. I do a lot of work in the browser and usually have 10+ tabs open in my system.
I also like to use the multi-window (Workspace) and have 8 Workspaces (thus one of the reasons for ccsm).
Once I leave the computer and try to come back (before the screen locks), the computer is COMPLETELY frozen with my HDD running at very high speeds. I have a system monitor that monitors the CPU activity and HD activity and right before those things freezes, my RAM is at 100% and so is my HDD.
I'm wondering if anyone can help me FIGURE OUT what the problem is? I've been using ubuntu since '08 so I'm familiar with it. I just don't know where to start!
HELP!
I can not deal with the problem until I install the following packages as follows
sudo apt-get install linux-generic-lts-quantal xserver-xorg-lts-quantal
You could get the detailed reason from http://www.howtoeverything.net/linux/hardware/random-freezes-integrated-hd-4000-graphics
Maybe you have to consider the Google Chrome or Chromium, whose two or more different Flash plugins conflict with each other, and the freeze problem disappear after you Disable the Flash plugins in the webpage opened by inputting
chrome://plugins
in the URL address of Chrome
Or another useful method of changing the chrome plugins is to uninstall the google-chrome completely,and then install the current chrome with newest flash plugin.
After all the above are done, the system need restart.
Hope it can help. Good luck!
I've been using VMWare Player for ages now for both Windows development on my Linux box and (more importantly) automated testing of Windows applications.
Basically what I do is to:
have my development VM running and I build my code and automatically transfer the install package to Linux.
when this shows up at Linux, automatically copy a "known-state", snapshot VM to my test work area (I say snapshot but it's really just a backup copy of the whole directory, not a real VMWare snapshot).
also automatically start the VM in the work area once it's copied.
the VM has a single never-changing startup script which pulls a real startup script from Linux and runs it.
that startup script is responsible for getting down the install package and doing a silent install.
it then runs a test suite and uploads results back to Linux where I have automated scripts which check them.
So, it's basically a one-button test process.
Now I notice more and more people seem to be using VirtualBox.
First off, I'd like to confirm that it can also do a similar thing, primarily being able to backup and restore whole VMs and having shared folders between VirtualBox and Linux.
Secondly, and this is the crux: I'd like to know if that has any concrete advantages over VMWare Player, especially for the automated testing jobs.
I switched to VirtualBox because of one concrete advantage, I wasn't able to setup the network as I wanted to in player. I don't remember if it was bridging or port-forward or whatever that didn't work, but something didn't work the way I wanted it to with the network-setup (cause I needed the pay-version for that) and thus I switched. Personally I've found that both have good and bad sides, but I still use virtualbox cause of that network-thing.