Survive a dead local status URL, and go unavailable when the spa is not reporting #2

Merged
dfritz merged 3 commits from ops/local-status-fallback into main 2026-11-15 04:51:24 +00:00
7 changed files with 118 additions and 7 deletions
Showing only changes of commit 3ba61e9c0b - Show all commits

No files matched your search

+20
View File
@@ -87,6 +87,24 @@ class HotSpringSpa:
self._client.loop_stop()
self._client.disconnect()
def subscribe(self) -> None:
"""Start reading telemetry from the broker (also after a reconnect)."""
self._subscribe = True
if self._client and self._connected.is_set():
self._client.subscribe(STATUS_WILDCARD.format(root=self.root_topic), qos=1)
def unsubscribe(self) -> None:
"""Stop reading telemetry from the broker and drop what it delivered.
Used when the local ``/status`` source takes over, so a stale cloud
value can never shadow a fresh local one.
"""
self._subscribe = False
if self._client and self._connected.is_set():
self._client.unsubscribe(STATUS_WILDCARD.format(root=self.root_topic))
self.state = {}
self.merged = {}
@property
def connected(self) -> bool:
return bool(self._client) and self._connected.is_set()
@@ -107,6 +125,8 @@ class HotSpringSpa:
self._connected.set()
def _on_message(self, client, userdata, msg):
if not self._subscribe:
return
suffix = msg.topic[len(self.root_topic) + 1:]
try:
payload = json.loads(msg.payload.decode("utf-8", "replace"))
@@ -82,6 +82,34 @@ class HotSpringConfigFlow(ConfigFlow, domain=DOMAIN):
step_id="user", data_schema=_schema(user_input), errors=errors
)
async def async_step_reconfigure(
self, user_input: dict[str, Any] | None = None
) -> ConfigFlowResult:
"""Change the local ``/status`` URL without signing in again.
The dongle takes its address from DHCP, so this is the one setting that
goes stale on its own.
"""
entry = self._get_reconfigure_entry()
if user_input is not None:
url = (user_input.get(CONF_LOCAL_STATUS_URL) or "").strip()
return self.async_update_reload_and_abort(
entry, data_updates={CONF_LOCAL_STATUS_URL: url}
)
return self.async_show_form(
step_id="reconfigure",
data_schema=vol.Schema(
{
vol.Optional(
CONF_LOCAL_STATUS_URL,
description={
"suggested_value": entry.data.get(CONF_LOCAL_STATUS_URL, "")
},
): str,
}
),
)
async def _validate(self, data: dict[str, Any]) -> str | None:
"""Return None on success, or an error key."""
cert_dir = data[CONF_CERT_DIR]
+7 -1
View File
@@ -82,9 +82,15 @@ CONF_CERT_DIR = "cert_dir"
# Optional: the dongle's local ``/status`` URL (e.g. http://<spa-ip>/status).
# When set, telemetry is polled locally instead of subscribed over the cloud MQTT
# broker, so monitoring stays on-LAN and off the vendor cloud. Control still uses
# the cloud MQTT connection. Only reachable when HA is on the spa's LAN.
# the cloud MQTT connection. Only reachable when HA is on the spa's LAN. The
# dongle has no DNS name and takes its address from DHCP, so this URL goes stale
# when the lease changes; the coordinator then reads from the cloud until the
# URL is corrected (reconfigure the entry) or answers again.
CONF_LOCAL_STATUS_URL = "local_status_url"
LOCAL_POLL_INTERVAL = 30 # seconds
# Consecutive failed local polls before telemetry is read from the cloud instead.
# The local poll keeps running and takes over again as soon as it answers.
LOCAL_FAILURES_BEFORE_CLOUD = 4
# The vendor mutual-TLS client certificate is NOT shipped in this repo (it is
# Watkins material extractable from the app APK). Place the three PEM files here
+38 -3
View File
@@ -6,6 +6,9 @@ Read path:
* If ``local_status_url`` is configured — poll the dongle's local ``/status``
endpoint instead, keeping monitoring on-LAN and off the vendor cloud. The MQTT
connection is still opened (control is cloud-only) but not subscribed.
* If that local endpoint does not answer (the dongle's DHCP address changed, or
it is off the LAN) — read from the cloud until it does. A dead local URL must
never take the whole integration, and with it control, down.
"""
from __future__ import annotations
@@ -30,6 +33,7 @@ from .constants import (
DEFAULT_CERT_DIR,
DOMAIN,
KEY_FILE,
LOCAL_FAILURES_BEFORE_CLOUD,
LOCAL_POLL_INTERVAL,
)
@@ -45,6 +49,10 @@ class HotSpringCoordinator(DataUpdateCoordinator[dict]):
self.spa_info: Spa | None = None
self._local_url: str | None = entry.data.get(CONF_LOCAL_STATUS_URL) or None
self._unsub_poll = None
# True while telemetry comes from the local URL; False while the cloud
# subscription is standing in for it.
self._local_ok = False
self._local_failures = 0
def _cert_paths(self) -> tuple[str, str, str]:
d = self.entry.data.get(CONF_CERT_DIR, DEFAULT_CERT_DIR)
@@ -69,10 +77,22 @@ class HotSpringCoordinator(DataUpdateCoordinator[dict]):
self.spa_info = spas[0]
creds = cloud.fetch_mqtt_credentials()
spa = HotSpringSpa(self.spa_info.root_topic, creds, on_status=self._on_status)
# Local read source -> connect for control only (no MQTT subscribe).
spa.connect(subscribe=not local)
status = None
if local:
spa.ingest_status(fetch_local_status(local))
try:
status = fetch_local_status(local)
except HotSpringApiError as err:
_LOGGER.warning(
"Spa local status URL %s does not answer (%s); reading "
"telemetry from the cloud until it does",
local,
err,
)
self._local_ok = status is not None
# Local read source -> connect for control only (no MQTT subscribe).
spa.connect(subscribe=status is None)
if status is not None:
spa.ingest_status(status)
else:
spa.wait_for_state()
return spa
@@ -95,7 +115,22 @@ class HotSpringCoordinator(DataUpdateCoordinator[dict]):
)
except HotSpringApiError as err:
_LOGGER.debug("local /status poll failed: %s", err)
self._local_failures += 1
if self._local_ok and self._local_failures >= LOCAL_FAILURES_BEFORE_CLOUD:
self._local_ok = False
_LOGGER.warning(
"Spa local status URL %s stopped answering (%s); reading "
"telemetry from the cloud until it does",
self._local_url,
err,
)
await self.hass.async_add_executor_job(self.spa.subscribe)
return
self._local_failures = 0
if not self._local_ok:
self._local_ok = True
_LOGGER.info("Spa local status URL %s answers again", self._local_url)
await self.hass.async_add_executor_job(self.spa.unsubscribe)
self.spa.ingest_status(status)
self.async_set_updated_data(self.spa.merged)
+1 -1
View File
@@ -8,5 +8,5 @@
"iot_class": "cloud_push",
"integration_type": "hub",
"requirements": [],
"version": "0.1.0"
"version": "0.1.1"
}
+12 -1
View File
@@ -13,6 +13,16 @@
"data_description": {
"local_status_url": "If Home Assistant is on the spa's LAN, e.g. http://<spa-ip>/status, monitoring is polled locally instead of over the cloud; control still uses the cloud."
}
},
"reconfigure": {
"title": "Hot Spring Connected Spa",
"description": "Change where monitoring is read from. The spa's address comes from DHCP and can change; while the URL does not answer, monitoring is read from the cloud.",
"data": {
"local_status_url": "Local /status URL (optional)"
},
"data_description": {
"local_status_url": "e.g. http://<spa-ip>/status. Leave blank to read from the cloud."
}
}
},
"error": {
@@ -22,7 +32,8 @@
"unknown": "Unexpected error."
},
"abort": {
"already_configured": "This spa account is already configured."
"already_configured": "This spa account is already configured.",
"reconfigure_successful": "Monitoring source updated."
}
},
"entity": {
@@ -13,6 +13,16 @@
"data_description": {
"local_status_url": "If Home Assistant is on the spa's LAN, e.g. http://<spa-ip>/status, monitoring is polled locally instead of over the cloud; control still uses the cloud."
}
},
"reconfigure": {
"title": "Hot Spring Connected Spa",
"description": "Change where monitoring is read from. The spa's address comes from DHCP and can change; while the URL does not answer, monitoring is read from the cloud.",
"data": {
"local_status_url": "Local /status URL (optional)"
},
"data_description": {
"local_status_url": "e.g. http://<spa-ip>/status. Leave blank to read from the cloud."
}
}
},
"error": {
@@ -22,7 +32,8 @@
"unknown": "Unexpected error."
},
"abort": {
"already_configured": "This spa account is already configured."
"already_configured": "This spa account is already configured.",
"reconfigure_successful": "Monitoring source updated."
}
},
"entity": {