Test the connection, or watch the run fail at zero
A green Healthy badge in the Environments list is a snapshot from up to several minutes ago. The only thing that proves the connection works right now is clicking Test Connection.
Every environment in DMM Infinity is reached through the agent module installed in your Mendix application. The agent sends a periodic heartbeat and answers an active test on demand. The Environments list shows you the heartbeat. It does not, on its own, tell you whether the agent will answer when a pipeline tries to talk to it.
Why this happens
Heartbeats arrive on a schedule, not in real time. After a Mendix redeploy, a network change, or a Cloud instance restart, the agent can stop sending without DMM Infinity noticing immediately. The badge stays green until the missed-heartbeat threshold passes; in the meantime any run you launch against that environment will fail at the first request — usually before records start moving, sometimes mid-batch.
The environment detail panel exposes two timestamps for exactly this reason: Last Heartbeatand Last Active Test. Heartbeat tells you the agent checked in. Active Test tells you DMM Infinity got a round-trip just now.
What to do instead
Treat Test Connection as the canonical proof. It runs in under a second and updates both timestamps on the spot. Build it into your habits, especially before any run that matters.
- ·Click Test Connection on the source and target environment before a destructive or large run.
- ·Re-test after any Mendix redeploy that includes the agent module.
- ·Re-test if Last Heartbeat is older than your missed-heartbeat threshold.
- ·Watch Last Active Test update to "a few seconds ago" — that is the green light.
- ·Trust a Healthy badge alone. It is up to several minutes stale.
- ·Schedule large overnight runs without a same-day Test Connection.
- ·Assume "the run will tell you" — it will, but only after you have wasted setup time queuing it.
A concrete example
A 200,000-record copy from QA to a sandbox is queued for 02:00. At 01:55 the QA Mendix instance restarts as part of a routine maintenance window; the agent module restarts a few seconds later, but the WebSocket back to DMM Infinity is blocked by a corporate proxy that the new instance picked up. The Environments list keeps showing Healthy until 02:03.
At 02:00 the run starts, fails on the first agent call, and stops at 0 / 200,000. The team finds out at 09:15 the next morning. Total time wasted: 7 hours. The 30 seconds you would have spent clicking Test Connection before queuing the run — and seeing it fail — would have given the maintenance window time to settle and the run a chance to actually move data.