Why does my integration fail with "<!DOCTYPE html>", "Socket hang up", "No healthy upstream" or 502?

Short answer: These errors are usually temporary server-side failures. Rerun the failed run from Run history. If a step keeps failing, reconfigure it.

Why it happens

• Requests time out or a connection drops under peak load. The error shows up as a truncated HTML page ("Doctype"), "Socket hang up", "No healthy upstream" or "502 Bad Gateway".
• Many integration runs executing at the same time can add to the load.
• Actions are versioned. A step saved on an older version can stop working after a platform upgrade.
• A duplicated action or integration can carry stale configuration.

How to fix it

1. Click Explore > Managed by me > Integrations > Run history, open the failed run and click Rerun. To retry many runs, use bulk rerun, up to 100 runs at a time. For a likely cause, click AI Troubleshoot on the failed run.
2. If the integration editor won't load, refresh the page, try an incognito window, and check your network.
3. If a step started failing though you changed nothing, reconfigure it: switch the action to a different connector, switch it back, configure it again and test.
4. If a duplicated action misbehaves, refresh it, or delete it and add a fresh action instead of a copy.
5. Rerun the failed runs after you reconfigure.
6. If failures keep coming back, note how often they happen, which action fails, whether large attachments or bulk payloads are involved, and whether they cluster at peak hours.

If failures are frequent, contact Kissflow support with the integration URL, failed run IDs and timestamps. See run history.

To retry temporary failures such as timeouts, server connection failures and unavailable services without stepping in, turn on Auto retry in the integration header. It's off by default and retries the failed step up to four times within about three minutes.