Job Management

Job lifecycle

State	Description
created	Job saved but not yet submitted — the only state in which its config can still be edited
queued	Job submitted, waiting for a worker
running	Worker is executing the job
success	All stages completed successfully; results available
failed	One or more stages failed; check logs
deleted	Job removed via Delete Job

Note

There is no cancel or resume. A running job runs to success or failed, and a failed job cannot be restarted from where it stopped — fix the cause and submit a new job.

Creating a job

1. From the Dashboard, click a pipeline card
2. Click Create New Job
3. Fill in the form (inputs vary by pipeline — see each pipeline’s docs)
4. Click Submit

The form validates all required fields client-side before submission. If validation fails, errors appear inline below each field.

Input bundle preflight

Some pipelines check your input bundle in your browser, before it uploads, and tell you what the pipeline will make of it. The check reads only column headers and file layout — nothing is sent anywhere to perform it.

It currently applies to the Clustering Pipeline and TakaraBio to AIRR. Other pipelines upload exactly as they always have. The panel is also entitlement-gated, so it may not appear even on those two — ask an administrator if you want it enabled.

Each file in the bundle comes back as one of three outcomes:

Outcome	Meaning
Usable	The pipeline will use this file
Cannot be used	The pipeline will refuse to run
Will be silently skipped	The pipeline will drop this file and carry on

Caution

“Will be silently skipped” is the outcome to read carefully. These files look like data files but are missing a column the pipeline needs. The job will still finish and still report success — it will simply have been run on a subset of your data, with nothing in the results to say what was left out. This is the case the check exists to catch; a job that fails outright is at least obvious.

The panel lists near-misses first — files that carried some of a required column set but not all — rather than burying them among vendor files that were never meant to qualify (QC tables, per-clone reports, alignment dumps).

You are never blocked. If the check reports problems you can still upload, after confirming a prompt. Bear in mind the pipeline runs this same check server-side and will stop with an error, so a bundle flagged Cannot be used is very likely to fail — override it only if you believe the check is wrong.

Note

Only column headers and file layout are checked. A bundle that passes can still fail later on the contents of those columns.

Viewing job details

The Job Management page shows

Column	Description
Job ID	Numeric identifier for the job record
Job Name	User-supplied label (not shown for FASTQ and SRA Data pipelines)
Workflow ID	Run id for the underlying workflow — click the copy button to grab it when reporting a problem
Patient Label / Patient ID	Only shown when every job in the list carries a patient label
Status	Current job state (created, queued, running, success, failed)
Created At	Timestamp when the job record was created

A few things you can do from Actions

• View Run — open the job details page
• View Logs — see below
• Edit Config — modify job parameters; only enabled while the job is still created, i.e. not yet submitted
• Create Context — turn a run’s output into a skill you can query with an LLM. Preview — the feature is not live yet
• Sync Status — force a re-read of the job’s real state from the workflow engine. Available for every job that has actually run, terminal ones included; only created and deleted jobs have nothing to sync against
• Download Output — download result files; only enabled on successful jobs
• View Report — open the in-app report viewer; only for pipelines with reports enabled, and only on successful jobs
• Delete Job — removes the job after a confirmation prompt. This cannot be undone, and it does not cancel an in-progress run

Activity — all pipelines at once

Each pipeline’s job list only shows that pipeline’s runs. Activity, in the top navigation bar and on the Dashboard, shows every job you have run across all pipelines in a single list — useful for finding a run when you can’t remember which pipeline it belonged to, or for checking what is currently in flight.

The table is the same one described above, plus a Pipeline column identifying each job’s pipeline and linking to it. The Actions menu behaves identically.

Viewing logs

What View Logs offers depends on whether your organisation has the full-logs entitlement.

	Without the entitlement	With the entitlement
When View Logs appears	Only on failed jobs	On any job
What it opens	A summarised view listing just the failed stages’ error messages	The full task list, with per-task logs

If you need the detailed view and only see the error summary, ask an administrator to enable full logs for your organisation.