Run History¶
Every task and chain execution is recorded as a run, with its status, trigger, timings, the rows it wrote and the parameters it used; this page explains each run status.
Where to find runs¶
- Flow detail page > Run History: runs of one flow. Pick Day, Week, Month or All, and type in "Filter by keyword".
- Admin > Task Analyze > Monitor (Flow Scheduler admins): runs across environments. See Monitor.
| Column | Shows |
|---|---|
| Status | Run status, as stored (see Run statuses) |
| Trigger | How the run started (see Triggers) |
| Schedule | The schedule that fired it; blank for a direct Run Now |
| Owner | Who created the task or chain; blank for a direct Run Now |
| Triggered | When the Flow Scheduler handed the job to Tableau |
| Tableau Start | When Tableau began running the flow |
| Completed | When the run ended |
| Wait | Time between Triggered and Tableau Start (queueing on Tableau) |
| Run time | Time Tableau spent running the flow |
| Records | Rows written, summed across output steps; blank when unknown |
| Params | Opens "Parameters used" with the values the run was submitted with |
| Details | The error text, or who started the run |
Run statuses¶
The Status column shows the status name exactly as below.
| Status | Group | Meaning |
|---|---|---|
PENDING |
Active | Run created; the Flow Scheduler is about to submit it |
SUBMITTED |
Active | Sent to Tableau; the Flow Scheduler is identifying the Tableau background job, for up to vizportal.adopt.window.sec.default |
RUNNING |
Active | Tableau job identified; the Flow Scheduler is following it |
WAITING_FOR_MMS |
Active | The MetaMiner Server (MMS) could not be reached; the Flow Scheduler retries for up to mms.retry.window.minutes |
WAITING_FOR_TAB |
Active | MMS is up but cannot reach Tableau; the Flow Scheduler retries |
SERVER_DOWN |
Held | Tableau Server was unreachable when a scheduled or catch-up run was due. The run is held until Tableau is back. A manual run is never held; it ends START_FAILED. |
SUCCESS |
Finished | The flow finished and all selected outputs were written |
FAILED |
Finished | The flow failed on Tableau, at least one output step failed although others were written, or the run went past job.timeout.hours after Tableau started it |
START_FAILED |
Finished | Tableau refused or could not start the flow, or MMS stayed unreachable past the retry window |
CANCELLED |
Finished | Cancelled by a user or an API client |
SKIPPED |
Finished | Not run: the previous run of the task was still active, an earlier chain step failed, or a held run was dropped because the next scheduled run was near |
COALESCED |
Finished | Tableau did not start a second copy because the same flow was already running, or a held run was replaced by a catch-up run. No failure e-mail is sent. |
LOST_VISIBILITY |
Finished | The Flow Scheduler could no longer see a job it was following, for vizportal.lostvisibility.timeout.sec. Check the flow's run history in Tableau for the real outcome. |
FAILED and START_FAILED count toward automatic disable. LOST_VISIBILITY has its own, separate limit. Defaults for the settings named above are in the Configuration Reference.
Triggers¶
| Trigger | Meaning |
|---|---|
SCHEDULED |
Fired by the schedule |
MANUAL |
Started with Run Now or Run |
| Catch-up | A missed slot run after the Flow Scheduler started again, or a held run released after a Tableau outage |
RECOVERY |
A replacement run, started after a Flow Scheduler restart because the earlier run's start on Tableau was not confirmed |
How a run is tracked¶
- The Flow Scheduler submits the flow through MMS and identifies the Tableau background job.
- It follows the job until it ends. When the environment has the Tableau repository configured in MetaMiner Administrator (MMA), it reads job status from the repository.
- At the end it records the result per output step and the rows written.
A run waiting in Tableau's queue does not time out; job.timeout.hours counts from the Tableau start.
Cancel a run¶
Flow Scheduler admins, the user who started the run, and the task owner can cancel it.
- Click Cancel on an active run in Run History, or select runs in Monitor and choose Cancel runs....
- Optional: type a Reason (optional).
- Click Cancel run. Keep running closes the dialog without cancelling.
The message reads "Run cancelled." or "Cancel sent to Tableau; the run stops in a few seconds." The status becomes
CANCELLED.
A run that has not reached Tableau yet is cancelled at once. Otherwise the cancel is sent once the Tableau job exists; after 30 minutes without a job, the Flow Scheduler stops trying. A cancelled chain stops before its next step.
To run again, click Run on the task row, or Run Now on the flow. See Tasks.
Clear or purge history¶
- Clear History on the flow detail page (Flow Scheduler admins) removes the flow's run history after asking for a reason.
- Delete history... in Monitor removes the selected runs.
- Run history is kept until you turn on automatic purge with
runs.retention.enabled. Then the leader node removes runs older thanruns.retention.daysonce a day, and only while a full backup on the leader node is younger thanruns.retention.max.backup.age.days. See Backup and Restore. - HA events, the Run Now queue, load history and e-mail records are pruned on their own schedules. See Configuration Reference.