HTCondor

HTCondor is a self-hosted workload manager for compute-intensive jobs distributed across networked machines. It suits organizations with technical capacity to manage their own computing resources. Its job recovery and configurable resource matching are strong reasons to consider it, though it is not a managed service.

Overview

HTCondor queues and schedules jobs, assigns priorities, monitors resources, and manages execution across clusters, cloud resources, and international grids. That breadth makes it relevant to workloads spread across varied computing resources rather than just a single workstation.

If a machine becomes unavailable during a job, HTCondor can restart the job elsewhere. It can also transfer job data when machines do not share a file system, or redirect input and output to the machine that submitted the job. These capabilities help keep distributed work moving, but they come with the responsibility of running a self-hosted system.

Key features

Scheduling and resource matching

HTCondor handles queues, priorities, scheduling, and resource monitoring. Its ClassAd mechanism compares job requests with machine offers using configurable requirements and preferences. That flexibility can help administrators direct jobs to suitable resources; it also makes HTCondor a better fit for teams able to define and manage those requirements than for users seeking a turnkey scheduler.

Recovery and data handling

Automatic restart on another machine can reduce disruption when a worker becomes unavailable. Data transfer and I/O redirection give jobs options when the submitting and execution machines lack a shared file system. Together, these features address practical obstacles in distributed execution, though they do not remove the need to operate the underlying installation.

Interoperability and monitoring

Flocking enables collaboration between HTCondor installations, while HTCondor-G interoperates with Globus-managed resources. The 25.14.1 feature release adds a metric daemon that can export metrics to Prometheus and Ganglia, giving operators integration options for monitoring.

Pricing

The HTCondor plan costs free. It includes the software, source code, and complete documentation under an open-source license. There is no free trial because the software is free, and deployment is self-hosted.

There are no paid tiers or seat and quota limits described. Contracted support is a separate option: it offers guaranteed problem resolution, faster issue turnaround, telephone access, and priority email access, with fees varying by support policy. Organizations that need that assurance should account for support fees; those running the free software rely on the terms of its license and their own operational capacity.

Platforms

HTCondor supports Linux, MacOS, and Windows, and is self-hosted. The range of supported operating systems offers flexibility for installations spanning different environments, while self-hosting means the organization remains responsible for deployment and operation.

Who it's for

HTCondor is aimed at organizations running compute-intensive work across shared or distributed resources. Its user base spans industry, government, and academia, with installations ranging from a handful to many thousands of workstations. That range suggests it can serve both modest and large deployments, but the self-hosted model makes in-house operational capacity an important consideration.

The project is housed within the Center for High Throughput Computing at the University of Wisconsin–Madison and receives NSF funding through the PATh project. HTCSS is licensed under Apache License, Version 2.0 and distributed without warranties or conditions. The project provides security fixes for the current feature release and current LTS release, and supports the previous LTS for approximately twelve months after a new LTS release. The current feature release is 25.14.1, dated September 24, 2026.

Pros and cons

• Pro: Automatic job restart on another machine can help recover work when a machine becomes unavailable.
• Pro: Configurable ClassAd matching lets administrators express requirements and preferences for assigning jobs to machines.
• Pro: Data transfer and I/O redirection support jobs that run without a shared file system.
• Con: Self-hosted deployment places operation and resource management on the organization, so it is not the choice for teams seeking a managed scheduler.
• Con: The software is provided without warranties or conditions; organizations wanting guaranteed resolution and faster support need to consider contracted support fees.

Alternatives

Job Scheduler Software is the broader category to explore when comparing schedulers.

• AWS Batch may suit teams that want to use AWS Batch without an additional service charge; compute and storage resources are billed separately, and AWS resource charges apply. It has a free plan and supports API, Linux, MacOS, web, and Windows.
• JS7 JobScheduler is another free-plan option, with API, Linux, MacOS, self-hosted, web, and Windows platforms. Its Open Source License is free under GPLv3, but high-availability clustering is not included.
• OpenPBS is a free, self-hosted alternative for Linux, MacOS, and API environments. Its open-source edition uses AGPL 3.0, and community forum support has no guarantees.
• PowerJob is a free open-source framework under Apache License 2.0, with API, self-hosted, and web platforms, and a free trial.
• Slurm Workload Manager is a no-monetary-cost, self-hosted cluster option under GNU GPL v2, with API, Linux, and self-hosted platforms.
• BMC Helix AIOps is a paid option with no free plan; pricing is by sales contact.
• ActiveBatch is a paid alternative with a free trial and quote-based pricing.

Verdict

Choose HTCondor if your organization needs to schedule compute-intensive jobs across distributed resources and can operate a self-hosted system. Its recovery, flexible resource matching, and data-handling options make it a capable fit for that work at no software charge. Look elsewhere if you need managed operations or guaranteed support without a separate support contract.