Stop reconnecting a session a dark appliance never opened
A switched-off washer or dryer fails in the DTLS handshake, not in a poll: `_poll_once` opens the session itself, so `_connect_session` runs to its 12s timeout with nothing to show. The poll path treated that like any other poll failure and ran its reconnect -- close the session, pause, poll again -- but there is no session to close and no association for the device to clean up, so the retry was the identical handshake five seconds later. That cost 29s of every 30s interval, and the same again on every setup attempt for an entry with no snapshot to load from. `_poll_once` now records which of the two failed, and the poll path skips the retry when the handshake is what never completed. A session that opened and then broke still reconnects within the cycle. The log was the half the reporters saw: an ERROR every cycle (plus a WARNING once three "reconnects" piled up) for a state this integration is built to sit through, which issue #269's reporter read as the integration having failed. An outage now reports once, DEBUG for the cycles after it, and INFO when the device answers again. Fixes #269
This commit is contained in:
3 files changed
+197
-21
No files matched your search
@@ -152,6 +152,32 @@ is keyed on `(domain, issue_id)`, `dataclasses.replace` in
|
||||
reloads non-persistent issues with their dismissal intact, so one row per
|
||||
entry survives restarts and an "Ignore" sticks.
|
||||
|
||||
### What a cycle costs while the appliance stays dark (issue #269)
|
||||
|
||||
An appliance switched off at the wall isn't a one-cycle blip: it fails the
|
||||
same way every 30s for hours, and both halves of that failure were being paid
|
||||
twice.
|
||||
|
||||
`_poll_once` opens the session itself when there isn't one, so a switched-off
|
||||
appliance fails *in the handshake* — 12s (`DtlsCoapSession.HANDSHAKE_TIMEOUT_S`)
|
||||
with nothing to show for it. The poll path then treated that like any other
|
||||
poll failure and ran its reconnect: close the session, pause
|
||||
`_RECONNECT_PAUSE_S`, poll again. There is no session to close and no
|
||||
association for the device to clean up, so the "reconnect" was the identical
|
||||
handshake five seconds later — 29s of the 30s interval spent proving the
|
||||
appliance is off, twice over, and the same again on every `SETUP_RETRY`
|
||||
attempt for an entry with no snapshot to load from. `_handshake_failed` marks
|
||||
that case in `_poll_once` so the poll path can skip the retry; a session that
|
||||
opened and *then* broke still reconnects within the cycle.
|
||||
|
||||
The log was the half the reporters actually saw: `poll failed after
|
||||
reconnect` at ERROR every cycle, plus a `reconnect_is_frequent` WARNING once
|
||||
three piled up, for a state this integration is specifically built to sit
|
||||
through. Issue #269's reporter read that repetition as the integration having
|
||||
failed. It's one ERROR per outage now, DEBUG for the cycles after it, and one
|
||||
INFO when the device answers again — HA's own coordinator already logs the
|
||||
transition into and out of a failed update.
|
||||
|
||||
## What this still won't do
|
||||
|
||||
Entities will be present and `unavailable` — not showing their last values.
|
||||
|
||||
Reference in new issue
Block a user