Bump version to v1.1.1 — fix MIN/MOD all-zero readings until manual reload

Fixes a regression from v1.0.10 (#351) affecting every MIN and MOD-family
profile, whose input registers live entirely in the 3000 range.

The 3000-range skip-on-failure suppression added in v1.0.10 assumed the block
was optional. On these profiles it is the only input range, so one failed read
suppressed all reads for 300s, left _register_cache empty, and every field
decoded to 0. read_all_data() returned that zeroed result rather than None, so
the coordinator saw a successful poll: entities stayed available showing 0,
_consecutive_failures never incremented, and no reconnect or adaptive backoff
ran. The window re-armed every 300s and persisted until a manual reload — most
visible as the overnight Waiting->Normal transition never being picked up.

- Skip suppression is no longer applied when the 3000 range is a profile's
  primary (only) input range; multi-inverter setups keep anti-log-flood behaviour
- Total failure of a primary range returns None so the coordinator marks the
  inverter offline and runs its reconnect/backoff path
- Universal guard: an empty register cache after all reads reports the poll as
  failed instead of publishing zeros as valid data
- Per-block success logging moved from INFO to DEBUG

Behaviour change: on genuine comms failure sensors now go unavailable rather
than reading 0 — correct HA semantics, since statistics ignore unavailable but
record 0 as a real measurement.

Also ships the MOD DO1.0 SOC control fix (#343, commit 6370a2b): holding
registers 3048 and 3067 replace the dead 1091/1071 pair.

Co-Authored-By: Claude <noreply@anthropic.com>

Author: 0xAHA

README.md

@@ -3,7 +3,7 @@
 # Growatt Modbus Integration for Home Assistant ☀️
 
 ![HACS Badge](https://img.shields.io/badge/HACS-Custom-orange.svg)
-![Version](https://img.shields.io/badge/Version-1.1.0-blue.svg)
+![Version](https://img.shields.io/badge/Version-1.1.1-blue.svg)
 [![GitHub Issues](https://img.shields.io/github/issues/0xAHA/Growatt_ModbusTCP.svg)](https://github.com/0xAHA/Growatt_ModbusTCP/issues)
 [![GitHub Stars](https://img.shields.io/github/stars/0xAHA/Growatt_ModbusTCP.svg?style=social)](https://github.com/0xAHA/Growatt_ModbusTCP)
 

RELEASENOTES.md

@@ -4,6 +4,47 @@
 
 ---
 
+## v1.1.1
+
+Issues: #343
+
+- **Fix: MIN/MOD inverters stuck reporting all zeros until manual reload (regression in v1.0.10):**
+  On profiles whose input registers live entirely in the 3000 range — every MIN and
+  MOD-family profile — a single failed read of that block put it into the 5-minute
+  "optional range" suppression window introduced in v1.0.10 (#351). Because the 3000 range
+  is the *only* input range on these profiles, suppression left the register cache empty
+  and every value decoded to 0. `read_all_data()` returned that zeroed result instead of
+  `None`, so the coordinator treated the poll as successful: entities stayed *available*
+  showing 0, the failure counter never incremented, and no reconnect or adaptive backoff
+  ever ran. The state re-armed every 300 s and persisted indefinitely — only reloading the
+  integration recovered it. Most visible as the overnight Waiting → Normal transition not
+  being picked up in the morning.
+
+  Three changes:
+  - Retry suppression is no longer applied when the 3000 range is a profile's primary
+    (only) input range. Multi-inverter setups keep the anti-log-flood behaviour.
+  - A total failure of a primary range now returns `None`, so the coordinator marks the
+    inverter offline and runs its reconnect/backoff path.
+  - Added a universal guard: an empty register cache after all reads reports the poll as
+    failed rather than publishing zeros as valid data.
+
+  **Behaviour change:** on a genuine communication failure, sensors now go *unavailable*
+  rather than reading 0. This is the correct Home Assistant semantic — the statistics
+  engine ignores unavailable states but records 0 as a real measurement, which previously
+  polluted energy history with false zeros.
+
+- **Fix: MOD charge/discharge SOC controls on DO1.0 firmware (#343):**
+  Added holding registers 3048 (`batt_first_charge_stopped_soc`) and 3067
+  (`grid_first_discharge_stopped_soc`) to the MOD 6000-15000TL3-XH profile. Registers
+  1091 and 1071 are dead on DO1.0 firmware — writes are accepted but have no effect,
+  and the whole 1060-1099 range reads back zeros. The 3000-range equivalents work
+  correctly. Confirmed by @Rohde2026 and @TimOsth.
+
+- Per-block "Successfully read N registers" logging moved from INFO to DEBUG. It fired on
+  every register block of every poll, burying genuine warnings in the HA log.
+
+---
+
 ## v1.1.0
 
 Issues: #322

custom_components/growatt_modbus/growatt_modbus.py

@@ -963,7 +963,7 @@ def read_input_registers(self, start_address: int, count: int, log_errors: bool
                     )
                     self._track_read_failure()
                     return None
-                logger.info("Successfully read %d registers from %d", len(registers), start_address)
+                logger.debug("Successfully read %d registers from %d", len(registers), start_address)
                 self._track_read_success()
                 return registers
 
@@ -1346,12 +1346,23 @@ def _read_sparse(addr_list: list, fatal: bool = False) -> bool:
             _3000_key = ('3000_block', count_3000)
             _3000_RETRY_S = 300
 
+            # The 3000 range is "optional" only for profiles that ALSO read a base range
+            # (0-874). For MIN/MOD-family profiles it is the only input range defined, so a
+            # failure here means the poll produced no data at all. Suppressing retries in
+            # that case leaves _register_cache empty for 5 minutes, every field below decodes
+            # to 0, and the coordinator publishes an all-zero reading as if the inverter were
+            # healthy — sensors read 0 instead of going unavailable, and none of the
+            # reconnect/backoff paths run because the poll "succeeded" (Issue: overnight
+            # Waiting->Normal transition never picked up until a manual reload).
+            _3000_is_primary = not has_base_range
+
             # Skip-on-failure: after the first failure, suppress retries for 5 minutes.
             # This prevents log flooding when an inverter simply doesn't support this range
             # (e.g. one model in a two-inverter setup where only one uses 3000-range registers).
+            # Never applied when the range is primary — see above.
             _3000_prev = self._failed_optional_ranges.get(_3000_key)
             _3000_skip = False
-            if _3000_prev:
+            if _3000_prev and not _3000_is_primary:
                 _3000_fail_time, _3000_fail_count = _3000_prev
                 if time.time() - _3000_fail_time < _3000_RETRY_S:
                     _3000_skip = True
@@ -1422,6 +1433,18 @@ def _read_sparse(addr_list: list, fatal: bool = False) -> bool:
                             self._register_cache[3000 + i] = value
 
                 # Update failure tracking based on outcome
+                if _3000_any_fail and not _3000_any_ok and _3000_is_primary:
+                    # Primary range failed completely — there is no data to report. Return
+                    # None so the coordinator marks the inverter offline, increments its
+                    # failure counter and runs its reconnect/backoff path, rather than
+                    # publishing zeros as a healthy reading.
+                    logger.warning(
+                        "Failed to read primary 3000 register block (3000-%d) — "
+                        "reporting poll as failed",
+                        max_3000_addr
+                    )
+                    return None
+
                 if _3000_any_fail and not _3000_any_ok:
                     _prev = self._failed_optional_ranges.get(_3000_key)
                     _count = (_prev[1] + 1) if _prev else 1
@@ -1555,6 +1578,19 @@ def _read_sparse(addr_list: list, fatal: bool = False) -> bool:
                     for i, value in enumerate(registers):
                         self._register_cache[min_addr_block + i] = value
 
+        # Safety net: if every range failed or was suppressed the cache is empty and every
+        # field below would decode to 0. Returning that as valid data makes the coordinator
+        # read a dead link as a healthy inverter reporting zeros — entities stay "available"
+        # showing 0, _consecutive_failures never increments, and no reconnect or adaptive
+        # backoff ever runs, so the state persists until a manual reload.
+        if not self._register_cache:
+            logger.warning(
+                "[%s@%s] No registers read this poll (all ranges failed or suppressed) — "
+                "reporting poll as failed",
+                self.register_map['name'], self.connection_id
+            )
+            return None
+
         # Now extract values using the register map
         try:
             # Status — look up by name rather than assuming it's at min_addr

custom_components/growatt_modbus/manifest.json

@@ -12,5 +12,5 @@
     "pymodbus>=3.0.0",
     "pyserial>=3.4"
   ],
-  "version": "1.1.0"
+  "version": "1.1.1"
 }