Talapas HPC Automation
Workflow automation for Talapas, the University of Oregon's heterogeneous HPC cluster — a Python tool that pulls pending account requests from the RACS Jira portal and routes each to the right PI for approval.
Overview
Work from my role as Student Administrator at Research Advanced Computer Services
(RACS), the group that runs Talapas — the
University of Oregon's heterogeneous high-performance computing cluster. This is the
infrastructure side of research: the tooling that keeps a shared cluster usable for the
labs that depend on it. (Source lives under RACS/ in the repo.)
Account-request automation
New Talapas users request accounts through a portal backed by Jira, and each request needs sign-off from the PI of the research group (PIRG) the user claims. Doing that by hand means reading tickets, looking up PIs, and writing emails — one at a time.
automated_account_requests.py automates the loop: it pulls pending account requests in
bulk from the portal via the Jira API, resolves each requester to their PIRG, and emails
the corresponding PI for approval. The script is deliberately environment-coupled — it
assumes Talapas's directory structure and module setup — because it's an operational
tool, not a library.
To be continued
The repo is the growing home for cluster automation and utility scripts from the RACS role; this node will pick up new tools as they land.