qtop¶
qtop is a fast text-mode utility for monitoring cluster utilization and scheduler state. It started as a shell script and is now actively maintained as a Python implementation.
This project follows a simple design goal: show the most useful cluster status information in one compact terminal view.
What qtop shows¶
qtop output is organized into three sections:
- Accounting summary
- Nodes: total, up, free
- Cores: available vs total
- Jobs: running and queued
-
Queues with running/queued counters
-
Worker node occupancy
- Node identifier (real name or remapped ID)
- Node state (
j,b,o,d,-for free) - Queue initials per node
-
Core-by-core job allocation matrix
-
User account information
- Single-character user ID marker
- Running vs total jobs
- Unix account name
- Optional GECOS / Grid mapping context
Why people use qtop¶
- Quickly assess LRMS health and utilization
- Track scheduler behavior over time (especially with watch mode)
- Spot unusual node behavior (black holes, imbalance, swapping patterns)
- Correlate users/jobs/nodes during incident response
- Export snapshots and replay state for troubleshooting
Scheduler support¶
The Python version is designed for multiple backends. In this repository, practical support and tests focus on the currently maintained integrations and demo mode.
Historically and conceptually, qtop covers systems from PBS/SGE families and has evolved to include additional schedulers through plugin work.
Project lineage¶
- Original concept and shell implementation: Fotis Georgatos
- Python rewrite and ongoing evolution: qtop contributors
The codebase in this repository is the Python translation and continuation of the original .sh tool.
Source:
fotis.web.cern.ch/fotis/QTOP
License¶
qtop is released under the MIT License. See LICENSE.