Compare commits

...

5 Commits

Author SHA1 Message Date
Jorijn Schrijvershof
81b7c6897a chore(main): release 0.2.7 (#25) 2026-01-06 09:54:44 +01:00
Jorijn Schrijvershof
a3015e2209 feat: add telemetry collection for companion and repeater nodes (#24)
Add environmental telemetry collection (temperature, humidity, barometric
pressure, voltage) from both the repeater node (over LoRa) and companion
node (local serial). Telemetry is stored in the same EAV metrics table
with `telemetry.` prefix.

Key changes:
- Add TELEMETRY_ENABLED feature flag (defaults to OFF)
- Add telemetry-specific timeout/retry settings
- Create shared telemetry.py module with extract_lpp_from_payload()
  and extract_telemetry_metrics() helpers
- Handle MeshCore API dict payload format: {'pubkey_pre': '...', 'lpp': [...]}
- Repeater: store status metrics BEFORE attempting telemetry (LoRa reliability)
- Companion: merge telemetry into single DB write (serial is reliable)
- Telemetry failures do NOT affect circuit breaker state

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-06 09:53:15 +01:00
Jorijn Schrijvershof
5545ce5b28 Merge pull request #23 from jorijn/release-please--branches--main--components--meshcore-stats
chore(main): release 0.2.6
2026-01-05 10:50:13 +01:00
Jorijn Schrijvershof
666ed4215f chore(main): release 0.2.6 2026-01-05 10:49:22 +01:00
Jorijn Schrijvershof
3d0d90304c fix: add tmpfs mount for fontconfig cache to fix read-only filesystem errors
The container runs with read_only: true for security hardening, but
fontconfig needs a writable cache directory. Added tmpfs mount at
/var/cache/fontconfig to allow fontconfig to write its cache without
compromising the read-only filesystem security.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 10:49:02 +01:00
11 changed files with 321 additions and 30 deletions

View File

@@ -1,3 +1,3 @@
{
".": "0.2.5"
".": "0.2.7"
}

View File

@@ -4,6 +4,20 @@ All notable changes to this project will be documented in this file.
This changelog is automatically generated by [release-please](https://github.com/googleapis/release-please) based on [Conventional Commits](https://www.conventionalcommits.org/).
## [0.2.7](https://github.com/jorijn/meshcore-stats/compare/v0.2.6...v0.2.7) (2026-01-06)
### Features
* add telemetry collection for companion and repeater nodes ([#24](https://github.com/jorijn/meshcore-stats/issues/24)) ([a3015e2](https://github.com/jorijn/meshcore-stats/commit/a3015e2209781bdd7c317fa992ced6afa19efe61))
## [0.2.6](https://github.com/jorijn/meshcore-stats/compare/v0.2.5...v0.2.6) (2026-01-05)
### Bug Fixes
* add tmpfs mount for fontconfig cache to fix read-only filesystem errors ([3d0d903](https://github.com/jorijn/meshcore-stats/commit/3d0d90304cec5ebcdb34935400de31afd62e258d))
## [0.2.5](https://github.com/jorijn/meshcore-stats/compare/v0.2.4...v0.2.5) (2026-01-05)

View File

@@ -16,6 +16,8 @@ Always edit the source templates, then regenerate with `python scripts/render_si
## Running Commands
**IMPORTANT: Always activate the virtual environment before running any Python commands.**
```bash
cd /path/to/meshcore-stats
source .venv/bin/activate
@@ -354,11 +356,17 @@ All configuration via `meshcore.conf` or environment variables. The config file
### Timeouts & Retry
- `REMOTE_TIMEOUT_S`: Minimum timeout for LoRa requests (default: 10)
- `REMOTE_RETRY_ATTEMPTS`: Number of retry attempts (default: 5)
- `REMOTE_RETRY_ATTEMPTS`: Number of retry attempts (default: 2)
- `REMOTE_RETRY_BACKOFF_S`: Seconds between retries (default: 4)
- `REMOTE_CB_FAILS`: Failures before circuit breaker opens (default: 6)
- `REMOTE_CB_COOLDOWN_S`: Circuit breaker cooldown (default: 3600)
### Telemetry Collection
- `TELEMETRY_ENABLED`: Enable environmental telemetry collection from repeater (0/1, default: 0)
- `TELEMETRY_TIMEOUT_S`: Timeout for telemetry requests (default: 10)
- `TELEMETRY_RETRY_ATTEMPTS`: Retry attempts for telemetry (default: 2)
- `TELEMETRY_RETRY_BACKOFF_S`: Backoff between telemetry retries (default: 4)
### Intervals
- `COMPANION_STEP`: Collection interval for companion (default: 60s)
- `REPEATER_STEP`: Collection interval for repeater (default: 900s / 15min)
@@ -410,6 +418,12 @@ Metrics are classified as either **gauge** or **counter** in `src/meshmon/metric
Counter metrics are converted to rates during chart rendering by calculating deltas between consecutive readings.
- **TELEMETRY**: Environmental sensor data (when `TELEMETRY_ENABLED=1`):
- Stored with `telemetry.` prefix: `telemetry.temperature.0`, `telemetry.humidity.0`, `telemetry.barometer.0`
- Channel number distinguishes multiple sensors of the same type
- Compound values (e.g., GPS) stored as: `telemetry.gps.0.latitude`, `telemetry.gps.0.longitude`
- Telemetry collection does NOT affect circuit breaker state
## Database Schema
Metrics are stored in a SQLite database at `data/state/metrics.db` with WAL mode enabled for concurrent access.

View File

@@ -15,7 +15,7 @@ services:
# MeshCore Stats - Data collection and rendering
# ==========================================================================
meshcore-stats:
image: ghcr.io/jorijn/meshcore-stats:0.2.5 # x-release-please-version
image: ghcr.io/jorijn/meshcore-stats:0.2.7 # x-release-please-version
container_name: meshcore-stats
restart: unless-stopped
@@ -47,6 +47,7 @@ services:
read_only: true
tmpfs:
- /tmp:noexec,nosuid,size=64m
- /var/cache/fontconfig:noexec,nosuid,size=4m
# Resource limits
deploy:

View File

@@ -102,6 +102,84 @@ Returns a single dict with all status fields.
---
## Telemetry Data
Environmental telemetry is requested via `req_telemetry_sync(contact)` and returns
Cayenne LPP formatted sensor data. This requires `TELEMETRY_ENABLED=1` and a sensor
board attached to the repeater.
### Payload Format
Both `req_telemetry_sync()` and `get_self_telemetry()` return a dict containing the
LPP data list and a public key prefix:
```python
{
'pubkey_pre': 'a5c14f5244d6',
'lpp': [
{'channel': 0, 'type': 'temperature', 'value': 23.5},
{'channel': 0, 'type': 'humidity', 'value': 45.2},
]
}
```
The `extract_lpp_from_payload()` helper in `src/meshmon/telemetry.py` handles
extracting the `lpp` list from this wrapper format.
### `req_telemetry_sync(contact)`
Returns sensor readings from a remote node in Cayenne LPP format:
```python
[
{'channel': 0, 'type': 'temperature', 'value': 23.5},
{'channel': 0, 'type': 'humidity', 'value': 45.2},
{'channel': 0, 'type': 'barometer', 'value': 1013.25},
{'channel': 1, 'type': 'gps', 'value': {'latitude': 51.5, 'longitude': -0.1, 'altitude': 10}},
]
```
**Common sensor types:**
| Type | Unit | Description |
|------|------|-------------|
| `temperature` | Celsius | Temperature reading |
| `humidity` | % | Relative humidity |
| `barometer` | hPa/mbar | Barometric pressure |
| `voltage` | V | Voltage reading |
| `gps` | compound | GPS with `latitude`, `longitude`, `altitude` |
**Stored as:**
- `telemetry.temperature.0` - Temperature on channel 0
- `telemetry.humidity.0` - Humidity on channel 0
- `telemetry.gps.1.latitude` - GPS latitude on channel 1
**Notes:**
- Requires environmental sensor board (BME280, BME680, etc.) on repeater
- Channel number distinguishes multiple sensors of the same type
- Not all repeaters have environmental sensors attached
- Telemetry collection does not affect circuit breaker state
- Telemetry failures are logged as warnings and do not block status collection
### `get_self_telemetry()`
Returns self telemetry from the companion node's attached sensors.
Same Cayenne LPP format as `req_telemetry_sync()`.
```python
[
{'channel': 0, 'type': 'temperature', 'value': 23.5},
{'channel': 0, 'type': 'humidity', 'value': 45.2},
]
```
**Notes:**
- Requires environmental sensor board attached to companion
- Returns empty list if no sensors attached
- Uses same format as repeater telemetry
---
## Derived Metrics
These are computed at query time, not stored:

View File

@@ -113,6 +113,23 @@ RADIO_CODING_RATE=CR8
# REMOTE_CB_FAILS=6
# REMOTE_CB_COOLDOWN_S=3600
# =============================================================================
# Telemetry Collection (Environmental Sensors)
# =============================================================================
# Enable telemetry collection from repeater's environmental sensors
# (temperature, humidity, barometric pressure, etc.)
# Requires sensor board attached to repeater (e.g., BME280, BME680)
# Default: 0 (disabled)
# TELEMETRY_ENABLED=1
# Telemetry-specific timeout and retry settings
# Defaults match status settings. Separate config allows tuning if telemetry
# proves problematic (e.g., firmware doesn't support it, sensor board missing).
# You can reduce these if telemetry collection is causing issues.
# TELEMETRY_TIMEOUT_S=10
# TELEMETRY_RETRY_ATTEMPTS=2
# TELEMETRY_RETRY_BACKOFF_S=4
# =============================================================================
# Paths (Native installation only)
# =============================================================================

View File

@@ -27,6 +27,7 @@ from meshmon.env import get_config
from meshmon import log
from meshmon.meshcore_client import connect_with_lock, run_command
from meshmon.db import init_db, insert_metrics
from meshmon.telemetry import extract_lpp_from_payload, extract_telemetry_metrics
async def collect_companion() -> int:
@@ -93,15 +94,26 @@ async def collect_companion() -> int:
else:
log.error(f"get_time failed: {err}")
# get_self_telemetry
# get_self_telemetry - collect environmental sensor data
# Note: The call happens regardless of telemetry_enabled for device query completeness,
# but we only extract and store metrics if the feature is enabled.
ok, evt_type, payload, err = await run_command(
mc, cmd.get_self_telemetry(), "get_self_telemetry"
)
if ok:
commands_succeeded += 1
log.debug(f"get_self_telemetry: {payload}")
# Extract and store telemetry if enabled
if cfg.telemetry_enabled:
lpp_data = extract_lpp_from_payload(payload)
if lpp_data is not None:
telemetry_metrics = extract_telemetry_metrics(lpp_data)
if telemetry_metrics:
metrics.update(telemetry_metrics)
log.debug(f"Extracted {len(telemetry_metrics)} telemetry metrics")
else:
log.error(f"get_self_telemetry failed: {err}")
# Debug level because not all devices have sensors attached - this is expected
log.debug(f"get_self_telemetry failed: {err}")
# get_custom_vars
ok, evt_type, payload, err = await run_command(
@@ -176,6 +188,10 @@ async def collect_companion() -> int:
summary_parts.append(f"rx={int(metrics['recv'])}")
if "sent" in metrics:
summary_parts.append(f"tx={int(metrics['sent'])}")
# Add telemetry count to summary if present
telemetry_count = sum(1 for k in metrics if k.startswith("telemetry."))
if telemetry_count > 0:
summary_parts.append(f"telem={telemetry_count}")
log.info(f"Companion: {', '.join(summary_parts)}")

View File

@@ -32,10 +32,10 @@ from meshmon.meshcore_client import (
get_contact_by_name,
get_contact_by_key_prefix,
extract_contact_info,
list_contacts_summary,
)
from meshmon.db import init_db, insert_metrics
from meshmon.retry import get_repeater_circuit_breaker, with_retries
from meshmon.telemetry import extract_lpp_from_payload, extract_telemetry_metrics
async def find_repeater_contact(mc: Any) -> Optional[Any]:
@@ -143,8 +143,10 @@ async def query_repeater_with_retry(
async def collect_repeater() -> int:
"""
Collect data from remote repeater node.
"""Collect data from remote repeater node.
Collects status metrics (battery, uptime, packet counters, etc.) and
optionally telemetry data (temperature, humidity, pressure) if enabled.
Returns:
Exit code (0 = success, 1 = error)
@@ -162,7 +164,8 @@ async def collect_repeater() -> int:
return 0
# Metrics to insert (firmware field names from req_status_sync)
metrics: dict[str, float] = {}
status_metrics: dict[str, float] = {}
telemetry_metrics: dict[str, float] = {}
node_name = "unknown"
status_ok = False
@@ -213,7 +216,7 @@ async def collect_repeater() -> int:
except Exception as e:
log.debug(f"Login not supported: {e}")
# Query status (using _sync version which returns payload directly)
# Phase 1: Status collection (affects circuit breaker)
# Use timeout=0 to let the device suggest timeout, with min_timeout as floor
log.debug("Querying repeater status...")
success, payload, err = await query_repeater_with_retry(
@@ -227,12 +230,12 @@ async def collect_repeater() -> int:
# Insert all numeric fields from status response
for key, value in payload.items():
if isinstance(value, (int, float)):
metrics[key] = float(value)
status_metrics[key] = float(value)
log.debug(f"req_status_sync: {payload}")
else:
log.warn(f"req_status_sync failed: {err}")
# Update circuit breaker
# Update circuit breaker based on status result
if status_ok:
cb.record_success()
log.debug("Circuit breaker: recorded success")
@@ -240,6 +243,51 @@ async def collect_repeater() -> int:
cb.record_failure(cfg.remote_cb_fails, cfg.remote_cb_cooldown_s)
log.debug(f"Circuit breaker: recorded failure ({cb.consecutive_failures}/{cfg.remote_cb_fails})")
# CRITICAL: Store status metrics immediately before attempting telemetry
# This ensures critical data is saved even if telemetry fails
if status_ok and status_metrics:
try:
inserted = insert_metrics(ts=ts, role="repeater", metrics=status_metrics)
log.debug(f"Stored {inserted} status metrics (ts={ts})")
except Exception as e:
log.error(f"Failed to store status metrics: {e}")
return 1
# Phase 2: Telemetry collection (does NOT affect circuit breaker)
if cfg.telemetry_enabled and status_ok:
log.debug("Querying repeater telemetry...")
try:
# Note: Telemetry uses its own retry settings and does NOT
# affect circuit breaker. Status success proves the link is up;
# telemetry failures are likely firmware/capability issues.
telem_success, telem_payload, telem_err = await with_retries(
lambda: cmd.req_telemetry_sync(
contact, timeout=0, min_timeout=cfg.telemetry_timeout_s
),
attempts=cfg.telemetry_retry_attempts,
backoff_s=cfg.telemetry_retry_backoff_s,
name="req_telemetry_sync",
)
if telem_success and telem_payload:
log.debug(f"req_telemetry_sync: {telem_payload}")
lpp_data = extract_lpp_from_payload(telem_payload)
if lpp_data is not None:
telemetry_metrics = extract_telemetry_metrics(lpp_data)
log.debug(f"Extracted {len(telemetry_metrics)} telemetry metrics")
# Store telemetry metrics
if telemetry_metrics:
try:
inserted = insert_metrics(ts=ts, role="repeater", metrics=telemetry_metrics)
log.debug(f"Stored {inserted} telemetry metrics")
except Exception as e:
log.warn(f"Failed to store telemetry metrics: {e}")
else:
log.warn(f"req_telemetry_sync failed: {telem_err}")
except Exception as e:
log.warn(f"Telemetry collection error (continuing): {e}")
except Exception as e:
log.error(f"Error during collection: {e}")
cb.record_failure(cfg.remote_cb_fails, cfg.remote_cb_cooldown_s)
@@ -248,28 +296,21 @@ async def collect_repeater() -> int:
# Print summary
summary_parts = [f"ts={ts}"]
if "bat" in metrics:
bat_v = metrics["bat"] / 1000.0
if "bat" in status_metrics:
bat_v = status_metrics["bat"] / 1000.0
summary_parts.append(f"bat={bat_v:.2f}V")
if "uptime" in metrics:
uptime_days = metrics["uptime"] // 86400
if "uptime" in status_metrics:
uptime_days = status_metrics["uptime"] // 86400
summary_parts.append(f"uptime={int(uptime_days)}d")
if "nb_recv" in metrics:
summary_parts.append(f"rx={int(metrics['nb_recv'])}")
if "nb_sent" in metrics:
summary_parts.append(f"tx={int(metrics['nb_sent'])}")
if "nb_recv" in status_metrics:
summary_parts.append(f"rx={int(status_metrics['nb_recv'])}")
if "nb_sent" in status_metrics:
summary_parts.append(f"tx={int(status_metrics['nb_sent'])}")
if telemetry_metrics:
summary_parts.append(f"telem={len(telemetry_metrics)}")
log.info(f"Repeater ({node_name}): {', '.join(summary_parts)}")
# Write metrics to database
if status_ok and metrics:
try:
inserted = insert_metrics(ts=ts, role="repeater", metrics=metrics)
log.debug(f"Inserted {inserted} metrics to database (ts={ts})")
except Exception as e:
log.error(f"Failed to write metrics to database: {e}")
return 1
return 0 if status_ok else 1

View File

@@ -1,3 +1,3 @@
"""MeshCore network monitoring library."""
__version__ = "0.2.5" # x-release-please-version
__version__ = "0.2.7" # x-release-please-version

View File

@@ -155,6 +155,14 @@ class Config:
self.remote_cb_fails = get_int("REMOTE_CB_FAILS", 6)
self.remote_cb_cooldown_s = get_int("REMOTE_CB_COOLDOWN_S", 3600)
# Telemetry collection (requires sensor board on repeater)
self.telemetry_enabled = get_bool("TELEMETRY_ENABLED", False)
# Separate settings allow tuning if telemetry proves problematic
# Defaults match status settings - tune down if needed
self.telemetry_timeout_s = get_int("TELEMETRY_TIMEOUT_S", 10)
self.telemetry_retry_attempts = get_int("TELEMETRY_RETRY_ATTEMPTS", 2)
self.telemetry_retry_backoff_s = get_int("TELEMETRY_RETRY_BACKOFF_S", 4)
# Paths (defaults are Docker container paths; native installs override via config)
self.state_dir = get_path("STATE_DIR", "/data/state")
self.out_dir = get_path("OUT_DIR", "/out")

102
src/meshmon/telemetry.py Normal file
View File

@@ -0,0 +1,102 @@
"""Telemetry data extraction from Cayenne LPP format."""
from typing import Any
from . import log
__all__ = ["extract_lpp_from_payload", "extract_telemetry_metrics"]
def extract_lpp_from_payload(payload: Any) -> list | None:
"""Extract LPP data list from telemetry payload.
Handles both formats returned by the MeshCore API:
- Dict format: {'pubkey_pre': '...', 'lpp': [...]}
- Direct list format: [...]
Args:
payload: Raw telemetry payload from get_self_telemetry() or req_telemetry_sync()
Returns:
The LPP data list, or None if not extractable.
"""
if payload is None:
return None
if isinstance(payload, dict):
lpp = payload.get("lpp")
if lpp is None:
log.debug("No 'lpp' key in telemetry payload dict")
return None
if not isinstance(lpp, list):
log.debug(f"Unexpected LPP data type in payload: {type(lpp).__name__}")
return None
return lpp
if isinstance(payload, list):
return payload
log.debug(f"Unexpected telemetry payload type: {type(payload).__name__}")
return None
def extract_telemetry_metrics(lpp_data: Any) -> dict[str, float]:
"""Extract numeric telemetry values from Cayenne LPP response.
Expected format:
[
{"type": "temperature", "channel": 0, "value": 23.5},
{"type": "gps", "channel": 1, "value": {"latitude": 51.5, "longitude": -0.1, "altitude": 10}}
]
Keys are formatted as:
- telemetry.{type}.{channel} for scalar values
- telemetry.{type}.{channel}.{subkey} for compound values (e.g., GPS)
Returns:
Dict mapping metric keys to float values. Invalid readings are skipped.
"""
if not isinstance(lpp_data, list):
log.warn(f"Expected list for LPP data, got {type(lpp_data).__name__}")
return {}
metrics: dict[str, float] = {}
for i, reading in enumerate(lpp_data):
if not isinstance(reading, dict):
log.debug(f"Skipping non-dict LPP reading at index {i}")
continue
sensor_type = reading.get("type")
if not isinstance(sensor_type, str) or not sensor_type.strip():
log.debug(f"Skipping reading with invalid type at index {i}")
continue
# Normalize sensor type for use as metric key component
sensor_type = sensor_type.strip().lower().replace(" ", "_")
channel = reading.get("channel", 0)
if not isinstance(channel, int):
channel = 0
value = reading.get("value")
base_key = f"telemetry.{sensor_type}.{channel}"
# Note: Check bool before int because bool is a subclass of int in Python.
# Some sensors may report digital on/off values as booleans.
if isinstance(value, bool):
metrics[base_key] = float(value)
elif isinstance(value, (int, float)):
metrics[base_key] = float(value)
elif isinstance(value, dict):
for subkey, subval in value.items():
if not isinstance(subkey, str):
continue
subkey_clean = subkey.strip().lower().replace(" ", "_")
if not subkey_clean:
continue
if isinstance(subval, bool):
metrics[f"{base_key}.{subkey_clean}"] = float(subval)
elif isinstance(subval, (int, float)):
metrics[f"{base_key}.{subkey_clean}"] = float(subval)
return metrics