Skip to content

Connection Health & Stale Connections

DataRecs continuously checks that your saved connections still work, so a broken connection is caught proactively — before a scheduled reconciliation silently fails against it.

About every 12 hours, DataRecs re-probes each of your connections with a lightweight connectivity check (the same check run when you press Test Connection). The result is shown on the connection’s Health tab as a status-strip of recent checks.

If a check fails, the retry timing depends on why it failed:

Failure kindWhat it usually meansDataRecs’ response
Invalid credentials / configurationRotated or wrong password, wrong host/port, TLS mismatchMarked stale immediately — retrying won’t help until you fix it
Timeout / network unreachableDatabase temporarily down, firewall/allowlist change, transient networkRetried after ~5 minutes, then ~1 hour; if still failing, marked stale

Any successful check resets the connection to healthy.

When a connection is marked stale, DataRecs protects your data and gets your attention:

  • Job schedules that use it are paused — so a broken connection doesn’t produce a stream of failed runs.
  • New runs are blocked — manual or scheduled runs that use the connection are refused (with a clear error) until it’s healthy.
  • We email the account admin and the person who created the connection, telling you which connection needs attention.
  • A webhook fires — the connection.check_failed event, if you have a subscription for it (see Using Webhooks). The payload tells you the connection_id and a leak-safe error_class — it never contains your credentials or connection string.
  1. Open the connection in the console (Connections → the connection → Health tab). The banner explains it’s stale and why.
  2. Fix the underlying problem — most often, click Edit and re-enter the rotated credentials, or restore network access to the database. Use Test Connection to confirm.
  3. Once the connection is healthy again, click Resume paused schedules.

If a schedule uses two connections and both went stale, fixing one connection won’t resume that schedule — it stays paused until both are healthy. When you resume, DataRecs tells you which schedules were resumed and which are still waiting on another stale connection, so there’s never any ambiguity about why a schedule is still paused.