TACC Job Manager is a lightweight python library for managing HPC resources provided by the Texas Advanced Computing Center (TACC).
TACCJM manages ssh connections to TACC systems for deploying applications, running jobs, and downloading/uploading data. These connections to are maintained by a locally deployed server that exposes an API to access TACC connections via http endpoints. This gives the user several methods to establish and maintain connections to TACC resources programmatically and without repeated 2-factor authentication. Furthermore the application and job hierarchy makes it easier to create reproducible and shareable HPC workflows for research.
Using taccjm requires a TACC account enable with 2-fa. In order to run jobs on TACC systems as well, you will need a valid allocation associated with your user ID enabled. See the TACC user portal
Warning
TACC Job Manager allows you to programmatically access TACC resources. Please be responsible in your access and usage of TACC resources. Review the TACC Usage Policy carefully before using any TACC resources.
To install use pip:
pip install taccjm
Alternatively the latest docker image can be pulled using
docker pull clos21/taccjm
The easiest way to use TACC Job Manager is through the Command Line Interface (CLI).
The CLI provides a useful set of help menus to navigate the commands available and how to use them.
Initializing a TACC connection requires logging in, but once logged, the connection is kept alive by the TACC Job Manager server that runs locally on your machine. Note the JM_ID is a user specified value (to be made an optional parameter in later versions).
Multiple connections to TACC resources can be established, however note that if more than one job manager is available, all subsequent commands will require an ID to be passed to specify which Job Manager needs to execute a command.
This project has been set up using PyScaffold 4.0.2. For details and usage information on PyScaffold see https://pyscaffold.org/.