Queue¶

The /queue/ page shows the current state of the docking queue across the whole server. Use it to see how busy PocketDock is before submitting a new job, or to track how your job is moving up the line.

What's on the page¶

Three sections, each auto-refreshed every 5 seconds:

Summary tiles at the top — three numbers:
- In queue — jobs waiting to start.
- Running now — jobs currently being processed by a worker.
- Done today — completed jobs in the last 24 hours.
Running now — jobs in any of the running stages (running_p2rank, running_prep, running_vina), with a live "running for X" timestamp.
Waiting — pending jobs in submission order. Each shows its position in the queue (#1, #2, …).
Recently finished — the last 50 jobs that finished in the last 24 hours, with their total duration.

Why names are redacted¶

Every row shows Job #<id> rather than the user-supplied job name or uploaded filenames. PocketDock's job URLs are public by default (anyone with the URL can see the results), so the queue page deliberately doesn't make it easier to discover what other people are docking. If you have your own job's URL — /jobs/<id>/ — you'll still see your own name and filenames there.

Estimated wait time¶

When your own job is pending, its status page shows two numbers:

Queue position — how many jobs are ahead of yours.
Estimated wait — when your job will start running (not when it'll finish).

The estimate is calculated as:

\[ \text{wait seconds} \approx \frac{(\text{jobs running} + \text{jobs queued ahead of you}) \times \text{avg duration}}{\text{worker concurrency}} \]

Where:

avg duration is the rolling average wall-clock of the last 20 completed jobs (defaults to 4 minutes when no history exists).
worker concurrency is the number of Celery worker processes — set via the WORKER_CONCURRENCY env var; should match --concurrency=N in docker-compose.yml.

Why the estimate may be off¶

The formula assumes all jobs take the same time and that running jobs just started. Reality is messier:

A job that's been running for 4 minutes will probably finish soon, but the formula still budgets a full average duration for it.
A job with exhaustiveness=32 takes ~4× longer than the default; the average doesn't know about per-job parameters.
P2Rank crashes that fail-fast pull the average down; large flexible ligands push it up.

Treat the ETA as a rough order-of-magnitude — useful for "should I wait or come back later" but not precise.

API¶

The same data is available as JSON at GET /api/queue/:

curl http://localhost:8000/api/queue/

Response shape:

{
  "pending_count": 3,
  "running_count": 2,
  "completed_today_count": 47,
  "worker_concurrency": 2,
  "avg_duration_seconds": 251,
  "running": [
    { "id": 142, "status": "running_vina", "status_display": "Running AutoDock Vina",
      "created_at": "2026-05-14T10:12:01Z", "updated_at": "2026-05-14T10:14:33Z" }
  ],
  "pending": [
    { "id": 143, "status": "pending", "status_display": "Pending",
      "created_at": "2026-05-14T10:14:00Z", "updated_at": "2026-05-14T10:14:00Z",
      "queue_position": 1 }
  ],
  "recent": [
    { "id": 141, "status": "completed", "status_display": "Completed",
      "created_at": "2026-05-14T10:08:01Z", "updated_at": "2026-05-14T10:11:42Z",
      "duration_seconds": 221 }
  ]
}

The same redaction applies — no name, no protein_file, no ligand_file.

Worker concurrency¶

PocketDock's Celery worker runs at --concurrency=2 by default (docker-compose.yml), meaning two jobs can run in parallel. To match the queue page's ETA calculation:

Setting	Where	Default
Celery process count	`--concurrency=N` in `docker-compose.yml`	`2`
ETA assumption	`WORKER_CONCURRENCY` env var	`2`

If you change one, change the other — otherwise the ETA on the queue page will be systematically wrong.