Schedd

Schedd is the scheduler daemon used in several distributed resource management systems to manage the submission, queuing, and dispatch of compute jobs. The term is most commonly associated with HTCondor and Sun Grid Engine (SGE), where a scheduler component coordinates how user jobs are accepted, scheduled, and run on available resources.

In HTCondor, the schedd runs on a submission host and handles user-submitted jobs. It reads job descriptions, creates internal representations, enqueues jobs, and applies policy decisions. It negotiates with execution resources through the system’s central services to find suitable machines (startd) for execution, and then dispatches jobs accordingly. The schedd tracks job state, handles holds, retries, and completion, and manages accounting data and logs. Administrators configure the schedd via configuration files and ClassAd expressions. The schedd communicates with the collector to publish its presence and status, and with the negotiator to obtain matches and trigger dispatch. User-facing tools such as condor_submit and condor_q interact with the schedd to submit jobs and query their status.

In Sun Grid Engine, the sge_schedd is the per-cluster scheduler daemon responsible for managing queues and dispatching jobs to compute hosts. It enforces user permissions, resource limits, and cluster policies as defined in the configuration, and interacts with the qmaster to maintain cluster state. When a job is submitted via qsub, the schedd places it in the appropriate queue, monitors its progress, and dispatches it to an execution daemon (execd) on a compute node. Users obtain status information through qstat and related tools, which reflect the scheduler’s view of the job lifecycle.

Overall, schedd denotes the core scheduler component in these systems, central to job submission, queuing, and resource dispatch across a computing cluster.