Hi all — filed this with support, who asked me to bring it to the forum for the engineering/product team. I’ve been collecting evidence for about a week with the beta deliberately kept installed. Two distinct failure signatures.
Setup
- Bridge: Bond Bridge Pro (
zermatt-pro-a), ID ZPGAIJ78296, firmware v4.34.1-beta (fw_date Jun 17), wired Ethernet, static IP, flat LAN (no VLANs) - Devices: 4 MagnaTrack RF units — Office SunShade (RMS98), Kitchen SunShade (RMS98), Dining Room Awning (RMS5), Awning Light
- BPUP client: Home Assistant Core 2026.6.4 (official
bondintegration). Two Bond app users exist but apps are normally closed — single effective BPUP subscriber.
Signature A — per-device push starvation (frequent: 4–15/device/day)
The bridge stops pushing BPUP state for one device for ~2 push cycles (~15–20 s) while pushing the other three normally in the same cycles. The client’s per-entity availability timeout trips; the next push (~1 s later) restores it. No errors, timeouts, or exceptions logged anywhere — the stream just goes quiet for one device.
Representative client-side capture (2026-06-28):
05:58:48 push: Dining, Kitchen, Office, Awning Light (all four)
05:58:58 push: Kitchen, Dining, Office (Awning Light MISSED)
05:59:08 push: Office, Dining (Awning Light still missing)
05:59:08 Awning Light -> unavailable (client availability timeout)
05:59:09 Awning Light -> available (next push arrived)
All four devices are affected, every day. Occasionally two devices starve in the same cycles (e.g. both SunShades at 2026-06-30 01:24:42 while Dining Awning + Awning Light kept pushing).
Signature B — full BPUP stream stall (rarer)
The entire BPUP stream stalls; the client’s keepalive gives up, tears down the UDP transport, reconnects and re-discovers. All four devices drop simultaneously.
Three verified events: 6/30 13:48:21, 6/30 16:17:34, 7/1 22:32:34 (PDT).
Methodology note: I screened every candidate event against six non-Bond devices on the same LAN and my hub’s MQTT connection log, and discarded any event within ±10 minutes of a Home Assistant restart — the three above are clean (client up, MQTT session connected throughout, no other device blipped).
Client log for the 7/1 22:32:34 event:
22:32:29 push: Dining, Office, Awning Light (normal cycle)
22:32:34 all 4 devices -> unavailable
22:32:35 [bond_async.bpup] BPUP connection stopping
22:32:35 version re-query: {'target': 'zermatt-pro-a', 'fw_ver': 'v4.34.1-beta', ...}
22:32:36 full device rediscovery (reconnect)
Ruled out
Wired bridge; no VLANs; no network changes during the observation window; non-Bond devices show zero availability blips at event times; no commands in flight (events hit at idle hours like 01:24 and 05:59); apps closed, so no client contention.
Impact
Not cosmetic: my wind/weather protective retract automations must refuse to act on unavailable/optimistic state, so a flap that coincides with a retract trigger silently skips the retract — a real problem for motorized awnings in gusts.
Happy to help
I have precise timestamps for dozens of events and can correlate with anything bridge-side. Glad to enable diagnostics, ship extended logs, stay on beta, or run test builds.
Bridge ID ZPGAIJ78296 · account pupEdog@gmail.com
