NOTING HERE IS TESTED
Install packages on Azure linux Virtual Machine instance. Install CUDA toolkit because you'll use this typically on a CUDA-Machine Save money by deallocating the VM instance after installation.
Run a [python] script with psub and forget. psub will deallocate resources once script is over.
- On your real local machine pull:
git pull https://github.com/pradyparanjpe/azure_scripts.git
-
Add your packages in the packages folder's copy inside your local machine
- Start Azure virtual machine and copy the contents of this modified git folder to the Azure virtual machine using a command such as
rsync -auvz ./azure_scripts [email protected]:~/.
Initiate azurecli Requirements
- See Requirements section
- Delete the folder azure_scripts
rm -rf ${HOME}/azure_scripts/
- remove line containing "azure_scripts" from the file
~/.bashrc
sed -i -e 's/^.*azure_scripts.*$//g' ${HOME}/.bashrc
- Run psub command of the form
nohup psub [-e <PACKAGE_FOLDER_NAME>] <command> &
- NOTE THE TRAILING '&'
- Kill a submitted process, and all child-process created by it.
p9kill -15 <PID>
- On Azure virtual machine,
psub linux_pypkg_install.sh [<PACKAGE_FOLDER_NAME>] 2>&1 >> installation.log &
After copying azure_scripts to the Azure VM,
-
Run linux_init_azurecli.sh with modifications
bash ~/azure_scripts/bin/linux_init_azurecli.sh -n <VMN> -g <VMG> -p <AZUREGITPATH>
- Login to azure virtual machine:
az login
-
Follow instuctions (Open a link in web-browser and supply the displayed
KEY
) -
Now, the AzureCLI API has been granted access to manage your Azure virtual machines.
- Update linux environment
source ~/.bashrc
- Python:
- Package should be pip3 compatible.
- setup.py must be present in the parent package folder, I do not look recursively.
- Only pip3 dependencies will be installed
- C/Cpp:
- Package should be generally installable by issuing
./configure; make; make install
- This feature is planned, and may be added in future.
- Package should be generally installable by issuing