Sync new rows¶
Your first Pipe already has one reading. Add another to its source:
mrsm sql local "INSERT INTO readings VALUES (2, '2026-01-01 13:00:00', 22.0)"
mrsm sync pipes -i sql:local -m temperatures
mrsm show data -i sql:local -m temperatures
The Pipe now has two readings: 21.5 and 22.0. You used the same sync command; the Pipe picked up the new row.
flowchart LR
A[New source rows] --> B[Sync the Pipe]
B --> C[Up-to-date table]Why the timestamp matters¶
A Pipe uses its latest timestamp to choose where the next fetch begins. It looks back over a small overlap to catch late arrivals, then compares the fetched rows with the table.
Already-seen rows stay put. New rows are inserted. Changed rows can update their matching keys.
What if data arrives late?
Increase the overlap with fetch.backtrack_minutes in the Pipe's parameters. For older history or a specific correction, sync an explicit window:
See Syncing for bounds, backtracking, and update options.
Watch a row arrive
Recorded with 5.0.0rc16. Playback pauses are extended for reading.
Keep it running¶
For a simple repeating sync, keep the command in the foreground:
Press Ctrl+C to stop. Later, a background job can give it a name and a schedule.