SARC stands for "Supervision et Analyde des Resources de Calcul". It's a Mila in-house tool to monitor the usage of computational resources by Mila users. It is not meant to be deployed on the users workstations.
This is the guide to use the client. For deployment, see docs/deployment.md.
To install sarc, git clone the repo and install sarc using poetry.
$ git clone [email protected]:mila-iqia/SARC.git
$ cd SARC
$ poetry installsarc will be looking into the current working directory to find the dev config file. To work with prod config or
from any directory set the environment variable as follow.
$ export SARC_CONFIG=/path/to/SARC/config/sarc-prod.jsonTo access the database, you need to setup the mila idt vpn and create an ssh tunnel. If you never accessed the VM, see documention here first https://mila-iqia.atlassian.net/wiki/spaces/IDT/pages/2216329289/VM+SARC.
To create the ssh tunnel:
$ ssh -L 27017:localhost:27017 sarcYou can now test on your machine a simple example to see if sarc is able to access the database:
$ poetry run python example/waste_stats.pyThose commands are for the proper formatting.
black .
isort --profile black .
tox -e isort
tox -e black
tox -e lint
TODO : How does poetry work? Needs a simple example here for the command to add a python module like ldap3. What are we supposed to type?
This runs the tests.
tox -e test
If you're running on Mac OS, you can install podman with brew install podman.
Later you can start the virtual machine with
podman machine init
podman machine start
sarc/account_matching/make_matches.py
sarc/account_matching/update_account_matches_in_database.py
sarc/inode_storage_scanner/get_diskusage.py (stub)
To generate documentation in HTML format in folder docs\_build:
sphinx-build -b html docs/ docs/_build
You can then open docs\_build\index.html on a web browser.