Agents on several accounts hit the same alarming message at the end of an
otherwise normal call:
ConnectionError (31005): Error sent from gateway in HANGUP
The call itself had already finished — audio, recording and call logging were
all fine — but the red banner made it look like something had gone wrong, and it
came back call after call. In at least one case it stopped a loan consultant from
working while three clients were trying to submit offers.
What’s Fixed
- No more error on a completed call. The calling service sends a
end-of-call notice as the call tears down; the dialer was displaying that
notice verbatim as a failure. It no longer does. - No more raw technical text. When something genuinely does go wrong, you
now get a message written for a human — “The call was declined”, for example —
instead of an internal error string and a number. - The dialer recovers on its own. If your calling session expires, it now
quietly renews in the background so your next call just works, rather than
leaving the dialer stuck until you reload the page. If the renewal itself
fails, you’ll be told plainly. - Calls that really do fail still tell you. A declined or failed call
shows a clear message as before — only the false alarm was removed.
Why it took a while to pin down
The message was reported from four different accounts, in two countries, on both
Sure Send-managed and customer-owned calling accounts — and each report looked
like a local network problem on that one person’s machine. It wasn’t. The
calling service was folding several genuinely different conditions into a single
generic error code, so every report arrived looking identical no matter what had
actually happened.
The dialer now asks the calling service for the specific reason instead of the
generic one, and records what it gets. That means the next time anything unusual
happens during a call, we can see how often it occurs and what actually caused
it — rather than waiting for someone to report it.