Every pipeline has a Settings tab where you control how it behaves: whether it retries failed records automatically, whether it runs continuously or on a schedule, what time range of data it brings in, and in what order it loads objects during the first sync. This page walks through each setting so you can configure a pipeline to match how your source and destination actually need to work.
Overview
The Settings tab groups its controls into five areas: Auto-Retry, Pipeline Schedule, Schedule configuration (when Scheduled is selected), Ingest Data, and Ingestion Order. The table below summarizes what each one does before the sections that follow go into detail.
| Setting | What it controls | Default |
| Retry Migration on Failure | Automatically retries failed record migrations to the destination | On |
| Pipeline Schedule | Runs the pipeline continuously (Real Time) or at set intervals (Scheduled) | Real Time |
| Schedule type | For Scheduled pipelines, run on a fixed Interval or a Cron Expression | Interval, every 15 minutes |
| Ingest Data | Which time range of source data the pipeline loads: historic, live, or both | Historic and Live Data |
| Ingestion Order | The order source objects load in during the historic sync | Not set |
Auto-Retry
The Retry Migration on Failure toggle tells the pipeline to automatically retry loading a record into the destination if that attempt fails. It's turned on by default, and you should generally leave it on. Turn it off only if you'd rather review and resolve failures manually from the pipeline's Logs tab instead of having the pipeline retry them on its own.
Real-Time vs. Scheduled Sync
The Pipeline Schedule setting determines when the pipeline moves data:
Real Time (default) — the pipeline ingests, transforms, and loads data to the destination continuously as changes happen at the source.
Scheduled — the pipeline runs at a recurring interval you define. Each run processes everything that's built up since the last run, then the pipeline goes idle until the next one starts.
Scheduled pipelines are easy to spot elsewhere in the product: the pipelines list shows their Ingestion as Scheduled, and the pipeline's own header displays "Scheduled for…" along with the time of the next run.
Configuring a Schedule
If you select Scheduled, choose a Schedule type:
Interval — set Trigger every to a whole number (default 15) and pick a Frequency of Seconds, Minutes (default), Hours, Days, or Weeks. You also set a required Starting at time and an optional Ending at time. Each frequency has a maximum interval: up to 59 for seconds and minutes, up to 23 for hours, and up to 29 for days.
Cron Expression — enter a standard cron expression, interpreted in UTC/GMT. The default is
0 0 * * *, which runs the pipeline once a day at midnight UTC.
Use Interval scheduling for straightforward, evenly spaced runs, and Cron Expression when you need a more specific pattern, such as running only on weekdays or at a particular hour.
Historic vs. Live Ingestion
The Ingest Data setting controls which time range of source data the pipeline brings in:
Historic and Live Data (default) — ingests both existing records and new changes going forward.
Live Data — ingests only data created from the moment the pipeline is set up onward; anything that existed before is skipped.
Historic Data — ingests only the existing, pre-existing data and does not pick up new changes afterward.
Not every source connector supports all three options — some sources only offer a subset, so the choices you see may be narrower depending on where the pipeline is pulling data from.
Ingestion Order
Ingestion Order is an optional setting for the historic (initial) load. It lets you control which source objects are ingested first, which is useful when objects depend on each other — for example, when one object references records that need to exist in another object first, such as with foreign key relationships.
Add objects to the Prioritised Objects list and drag them into the order you want. Objects in this list are ingested in the priority order you set, ahead of all other objects in the pipeline. This setting only affects the historic load — it has no effect on real-time changes, which continue to flow in as they occur regardless of ingestion order.
Publishing Your Changes
Settings changes don't take effect the moment you save them. Like object selections and schema changes, they're held as part of the pipeline's draft version. Once you save a change, the pipeline's header shows an Unpublished Changes pill as a reminder. Your updated settings only take effect after you deploy the new version from the pipeline.
Notes
Auto-Retry stays on by default — turning it off means failed records need manual attention from the Logs tab.
Switching a pipeline from Real Time to Scheduled changes how the pipelines list and header describe its status.
Interval limits differ by frequency: 59 for seconds and minutes, 23 for hours, 29 for days.
Cron expressions run in UTC/GMT, not your local time zone.
Available Ingest Data options can be narrower for some source connectors.
Ingestion Order only governs the historic load, not ongoing real-time updates.
Any setting you change stays in draft until you deploy the pipeline.