Last thing the SSO bridge (iserver/auth/ssodh/init) said, observed as
those calls pass through the proxy.
Why this is worth keeping: the Gateway can be ALIVE AND WRONG. On
2026-09-03 that endpoint returned HTTP 500 for eleven hours while
/health, auth/status and the process table all looked perfect —
meaning no login could complete, and the only symptom anyone could see was
2FA “not working”. A restart cleared it and the same call answered 401.
Recorded PASSIVELY rather than probed on a timer: the relogin service and
the watchdog already POST this endpoint through us, so watching what goes
by costs nothing and cannot add load to a Gateway that is already unwell.