Managing Jobs via the Cluster Monitor
This Cluster Monitor window shows summarised information on the selected workflows.
Note: The amount of data read and written is not available while a task is running, so the number is based on completed and errored tasks.
Note: IO (input/output, i.e. reading and writing) represents a scarce, shared resource on the file systems. Jobs that are particularly punishing to network disks (I/O intensive) are marked as such, and there is a limit as to how many of them can run concurrently.
The job table can give insight into whether a job queueing is I/O related.
Prio: Current job priority.
R: Shows the number of tasks that are running. I/O intensive jobs are indicated by a (*).
Q: Shows the number of tasks that are queued.
D: Shows the number of tasks that are dependent and awaiting completion of other tasks.
H: Shows the number of tasks that are on hold.
IO: Shows the number of jobs that are queued for I/O-related reasons. A (*) in this column indicates that the job is I/O intensive but is queuing for other reasons.
E: Shows the number of tasks that are errored.
These statuses are also reflected in the tasks table, under the Status column.
The following are job modification options to perform simple changes to jobs queueing on the cluster. If multiple tasks are selected, job modifications will only apply to applicable tasks.
Set Priority: Can only be applied to Queueing or Dependent jobs.
Set Partitions: Can only be applied to non-running jobs.
Hold: Places jobs into Held status, can only be applied to Queueing jobs.
WARNING: Selecting Requeue Hold, Requeue, Release, and Cancel Jobs options will result in the loss of any work that has been completed.
Requeue Hold: Places jobs into Held status, can only be applied to Running jobs.
Requeue: Restarts a job from scratch, can only be applied to Running jobs.
Release: Releases Errored or Held jobs into Queueing status.
Cancel Jobs: Deletes selected jobs from the cluster.