- Home
- GPU Cluster Management Software
- HTCondor
HTCondor
Madison, Wisconsin, United States
What it is
HTCondor creates a high-throughput computing environment that uses networked machines to run submitted jobs. It can distribute many jobs, including thousands of runs across multiple data sets, and use otherwise idle machines in a pool. ClassAds match each job’s requirements and preferences with resource offers from machines. If a machine becomes unavailable, HTCondor can restart the job elsewhere. Jobs can run without users having login accounts on execution machines through file transfer and split execution. The system can use clusters, cloud resources, and international grids, while flocking connects multiple HTCondor installations. File transfers to and from S3-compatible storage are supported, including Google Cloud Storage through its S3 interoperability API, but whole buckets or directories are not supported. HTCondor includes Python bindings, Kubernetes support, quota controls, GPU utilization metrics, and cloud GPU support. It is free under Apache License 2.0 and supports Linux, macOS, Windows, web, API, and self-hosted use. Community mailing-list support is free; fee-based contracts provide guaranteed resolution and faster turnaround.
For and against
For
- Schedules large batches across available machines
- Restarts jobs after machine unavailability
- No execution-machine login accounts required
- Supports clusters, clouds, and international grids
- Apache License 2.0
Against
- Whole S3 buckets or directories cannot transfer
- Older LTS channels eventually stop bug fixes
- Paid contract support is separate from community support
The verdict
HTCondor is a free scheduler for distributing and recovering large computational workloads across networked resources. Its matching model, transfer mechanisms, and cloud or grid options support varied self-hosted environments.
Who it suits
It suits academic, government, and commercial organizations managing distributed computational workloads across local, cloud, or grid resources.
On the record
- Access model
- HTCondor does not require users to have a login account on the machines where their jobs run, using file transfer and split execution mechanisms.htcondor.com · read 30 Sep 2026
- Audience
- The site says HTCSS is used by academic, government, and commercial organizations to manage computational workloads.htcondor.com · read 30 Sep 2026
- Cloud and grid use
- The overview says HTCondor can use clusters, cloud resources, or international grids, and the product page describes flocking multiple HTCondor installations together.htcondor.org · read 29 Sep 2026
- Cloud storage integration
- HTCondor supports file transfers to and from S3-compatible storage, including Google Cloud Storage through its S3 interoperability API, but does not support transferring entire buckets or directories.htcondor.readthedocs.io · read 29 Sep 2026
- Community support
- Free, community-based best-effort support is available through the htcondor-users mailing list.htcondor.org · read 29 Sep 2026
- Contract support
- Fee-based internal support offers guaranteed problem resolution, faster turnaround, telephone access, and a high-priority email address.htcondor.org · read 29 Sep 2026
- Documentation
- The site provides user and administrator quick-start guides and a manual that can be viewed online or downloaded.htcondor.com · read 30 Sep 2026
- Execution model
- HTCondor can run jobs without requiring users to have login accounts on the machines where the jobs run.htcondor.org · read 29 Sep 2026
- HTCondor-CE integration
- HTCondor-CE supports Grid Engine, HTCondor, LSF, PBS Pro, Torque, and Slurm as local batch systems.htcondor.com · read 30 Sep 2026
- Job scheduling
- Users submit jobs to HTCondor, which finds available machines and starts running the jobs there.htcondor.com · read 30 Sep 2026
- License
- HTCondor Software Suite is licensed under Apache License, Version 2.0.htcondor.org · read 29 Sep 2026
- Purpose
- HTCondor creates a High-Throughput Computing environment by using the computing power of networked machines, clusters, cloud resources, and international grids.htcondor.com · read 30 Sep 2026
- Python
- The site says HTCondor comes with Python bindings.htcondor.com · read 30 Sep 2026
- Recovery
- If a machine running a job becomes unavailable, HTCondor can automatically restart the job on another machine.htcondor.com · read 30 Sep 2026
- Release channels
- The download page describes feature releases for new features and bug fixes, and LTS releases for bug fixes to a fixed feature set.htcondor.com · read 30 Sep 2026
- Resource matching
- HTCondor uses ClassAds to match and rank resource offers against job requests, considering resource properties and priority values.htcondor.com · read 30 Sep 2026
- Scale
- The overview describes using HTCondor to submit many jobs at once, including thousands of runs across multiple data sets.htcondor.org · read 29 Sep 2026
- Security support
- HTCondor provides security fixes for the current feature version and current LTS version, and supports the previous LTS for approximately twelve months after a new LTS release.htcondor.com · read 30 Sep 2026
