Syncs
A sync is the process that keeps a connected data source's data up to date inside the context repository, on a schedule or a manual trigger.
Category
Product feature
See all terms
What is a sync?
A sync is the operation that pulls data from a connected source and keeps it up to date inside the context repository. Qontext runs four kinds: full_sync for a complete transfer, incremental_sync for delta-based updates, live_sync for continuous real-time updates, and manual_sync for a sync a user triggers directly. Every sync moves through the same phases: preparing, where it counts and analyzes what needs to move, data_fetching, where it retrieves the source data, and a final loading phase that writes the records into the repository.
Why syncs exist
Source data changes constantly, and none of it is useful to an agent if it only reflects what the source looked like on the day someone last connected it. A sync is what keeps a data source's copy in the context repository current without anyone re-importing it by hand. The first sync against a large dataset can take up to twelve hours, since it has to move everything selected for the first time; every sync after that only processes what changed, which is why incremental syncs finish far faster than the initial one.
How do syncs work?
Each sync is tracked as a sync log, tied to a pipeline and a workspace, with a status that moves through pending, preparing, data_fetching, and ingesting, on to completed, failed, stopped, or skipped. A sync log records when it started and completed, how many records it set out to sync against how many it actually synced, and if it failed, both a technical error message and a plain-language one shown in the interface. Filters set per data source control what a sync actually pulls in, so a sync only moves the data a workspace has chosen to bring in rather than everything the connected account can see. Every synced record becomes part of the same context repository that context agents read from and users query directly.