Skip to main content

Meshtastic part 2: MQTT is useful—but it is not a radio path

Beginner to intermediate · No server required to understand the concept

A node can reach its neighbor directly over LoRa, and other Meshtastic nodes may relay packets over more radio hops. An optional Internet-connected gateway can also move packets between radio communities using .

That combination is powerful. It is also easy to misunderstand: a long-distance message might have crossed the Internet rather than the air between those two communities.

Meshtastic local radio networks that can continue communicating independently of an optional MQTT Internet bridge.

First, what is an MQTT broker?​

An is a message-distribution server. A gateway can publish packets to a named topic, and a second gateway can subscribe to a topic and receive compatible packets. The broker usually uses an Internet or other IP network connection. It does not transmit LoRa RF by itself.

Think of it as two separate systems:

  • LoRa RF: Radio waves between compatible Meshtastic nodes, possibly with other nodes forwarding over radio.
  • MQTT/IP: Packets travel through your router, Wi-Fi/cellular service, an Internet connection and a broker, then perhaps back to another radio gateway.

A self-hosted or privately run broker changes who controls the server; it does not automatically remove the Internet dependency if gateways reach it across the public Internet.

Example: two separate Michigan communities​

A node in Clare County may reach a local MQTT-connected gateway over radio. That gateway may publish the packet to an MQTT broker. Another gateway somewhere else subscribes and may make selected packets available to its own radio neighborhood.

The distant node might see a message even though there is no over-the-air link between the communities. When judging radio coverage, separate RF reachability from broker reachability.

The two directions of a gateway​

DirectionWhat happensWhy it matters
Uplink (radio → MQTT)A local gateway hears a selected packet and publishes it to the brokerOff-site apps or other gateways can see it, depending on topic access and encryption
Downlink (MQTT → radio)A gateway receives a selected broker packet and makes it available locally over RF, subject to firmware and broker policiesAdds Internet-origin traffic to limited local airtime
NeitherNodes exchange RF packets without publishing or receiving broker trafficKeeps the local mesh independent of the broker

Uplink and downlink are configured at the channel level in addition to enabling the MQTT module. Enabling the module alone does not necessarily bridge any channel.

Northern Michigan Mesh broker settings​

For the independently operated Northern Michigan Mesh broker (not the default Meshtastic public broker), use these intentionally shared community credentials:

Meshtastic MQTT settingNorthern Michigan Mesh value
Server Addressmqtt.northernmichmesh.org
TLSEnabled
TCP port8883
Usernamemeshpublic
Passwordrogueregion
Root Topicmsh/US/MI/NMI
Recommended first testUplink only on a deliberately configured private channel; downlink off

Important address-field distinction: the MeshCore MQTT Observer uses the entire mqtts://mqtt.northernmichmesh.org:8883 URL in its Server field. The usual Meshtastic MQTT settings separate the hostname from the TLS setting (and may offer a separate port field). Do not assume that pasting a URL works in a hostname-only field.

The meshpublic / rogueregion pair is public intentionally, not an administrator login. Broker permissions still control what topics it can access, and credentials may change if abused. Do not use the default Meshtastic public-broker credentials or your server admin password on this broker. Begin with a private channel and uplink only. The fully illustrated setup and troubleshooting are available on Northern Michigan Mesh MQTT.

How to explore MQTT safely​

These are concepts to verify in the device's current app or firmware, not instructions to turn every switch on indiscriminately:

  1. Start from a working RF-only test. Confirm two nodes can communicate when MQTT is disabled.
  2. Select the gateway. It needs an IP connection, either through supported Wi-Fi/Ethernet hardware or a supported client proxy using a connected phone's Internet connection.
  3. Choose a broker. It might be the project's public broker or one administered by an independent group. Understand access credentials, logging, topic permissions and privacy policies first.
  4. Review the root topic. This is a namespace for published messages, not an RF frequency or a geographic transmit limit.
  5. Consider uplink only at first. It can help you observe traffic without immediately injecting Internet packets into local RF airtime.
  6. Only enable downlink deliberately after checking community airtime, server policies, keys and whether Internet messages should enter your local radio network.
  7. Review encryption and map reporting separately. “MQTT enabled,” “JSON enabled,” “TLS enabled” and “Map reporting enabled” are not equivalent privacy settings.
  8. Test disconnection. Temporarily remove the gateway's Internet path, not its radio power, and see which conversations still work.

Finding the controls in the current apps​

The official Meshtastic MQTT documentation currently shows the following routes:

ClientMQTT module settingsChannel uplink/downlink
AndroidSettings → MQTTSettings → Channels → select your intended channel
Apple / iOSSettings → Module Configuration → MQTTSettings → Channels / Primary Channel
Meshtastic Web ClientConfig → Module Config → MQTTChannels → selected channel

With Wi-Fi or Ethernet gateways, configure network access in the device's Network settings as supported. With compatible mobile apps, the MQTT client proxy can use the phone's Internet connection instead; the device still needs its local radio interface and power.

The settings to understand are Enable, Server Address, Username/Password, TLS, Root Topic, Encryption Enabled, JSON Enabled, Client Proxy, Map Reporting, and per-channel Uplink/Downlink. Not every device/client supports every combination. Review privacy and network policies before enabling.

Example: a controlled uplink-only test on a private channel you own using the Meshtastic Python CLI. Review the current supported syntax for your firmware and client first:

# Inspect the connected device and confirm your intended channel.
meshtastic --info

# Configure this independent broker and its public community credentials.
meshtastic --set mqtt.address mqtt.northernmichmesh.org
meshtastic --set mqtt.username meshpublic
meshtastic --set mqtt.password rogueregion
meshtastic --set mqtt.tls_enabled true
meshtastic --set mqtt.root msh/US/MI/NMI
meshtastic --set mqtt.encryption_enabled true
meshtastic --set mqtt.json_enabled false
meshtastic --set mqtt.map_reporting_enabled false
meshtastic --set mqtt.enabled true

# Example only: index 0 must be YOUR intended private channel.
# Do NOT use this blindly on the public LongFast channel.
meshtastic --ch-set uplink_enabled true --ch-index 0
meshtastic --ch-set downlink_enabled false --ch-index 0

Only the community MQTT password shown here is intended to be public. Private channel pre-shared keys, Wi-Fi passwords and server administration credentials must still remain private.

Start with downlink disabled, confirm published packets on a broker you're permitted to observe, and test your radio independently. Change only settings you understand; the gateway may reboot after configuration changes.

The official MQTT module configuration provides current app steps, CLI names and details. Menus and behavior may change with firmware/app releases; verify the current instructions before applying settings.

Client proxy isn't a universal feature

Some devices have no Wi-Fi/Ethernet and can use a connected phone as their Internet proxy in supported mobile applications. That phone must actually have connectivity while proxying; an offline phone does not make a functioning Internet bridge. The Meshtastic MQTT documentation notes that Web-client proxy behavior may not match a mobile client. Check your exact platform.

What if the Internet goes down?​

Imagine four nodes:

  • A and B can communicate by local LoRa radio.
  • C and D can communicate in another local LoRa neighborhood.
  • A gateway in each neighborhood connects through the Internet to the same MQTT broker.
FailureA ↔ B locallyC ↔ D locallyA ↔ C between areas
Everything workingWorks if RF path is goodWorks if RF path is goodMay work through MQTT, depending on configuration
Broker unavailableStill works by RFStill works by RFMQTT route fails
One gateway loses InternetStill works by RFStill works by RFMQTT route fails
Cellular service fails for phone-proxied gatewayStill works by RF, if local stations have powerStill works by RFPhone-proxied MQTT route fails
Gateway loses all powerRF paths not involving that node may workRF paths not involving that node may workMQTT path fails; RF relaying by that node also stops
Local RF path brokenMay failMay still workMQTT does not automatically repair missing local RF hops

These are simplified examples, not guaranteed message delivery. In real networks multiple gateways or alternate RF paths may change the outcome.

Most important lesson: if two communities have no radio path between them, their MQTT-assisted connection is an Internet-dependent link. Calling the entire link “off-grid” without that qualification can be misleading. Local Meshtastic radio still works without the Internet where a suitable RF path exists.

A locally hosted broker could remain reachable during a public Internet outage if gateways have a functioning independent local IP path to it. That still means the interconnection depends on that IP network, its power, and the broker—not solely on LoRa.

Public MQTT servers are not unlimited RF relays​

The official Meshtastic public MQTT service applies policies to protect performance. At the time of the reviewed documentation, those include zero-hop limits for public-broker-origin traffic, traffic filtering and location precision protections. That means a packet received from the public broker should not be assumed to flood across every connected local radio node.

Likewise, turning on multiple gateway downlinks can duplicate traffic and consume radio airtime. Different privately operated brokers may apply different rules.

Public keys and private brokers

Meshtastic's current integration documentation discourages using the default publicly shared channel key on a private broker. Without the public service's safeguards, broad uplink/downlink of public-channel traffic may create security risks or flood a local radio mesh. Prefer an intentional private channel with a suitable key, scoped topic access and carefully controlled downlink policies. A private hostname by itself does not make message contents private.

Read the official MQTT integration and public server policy.

Understanding topics—and avoiding a common mistake​

You might see MQTT topics that look something like:

msh/US/2/e/LongFast/!abcd1234

That's a topic path containing a region namespace, packet format portion, channel identifier and node identifier—not a proof of the route taken by a packet over radio. The exact root topic and packet format depend on settings and firmware. The integration docs describe protobuf messages, and optional JSON messages for external integrations.

Plain-English takeaway: Seeing a packet on an MQTT topic proves that a reporting or bridging system delivered it there; it does not prove your antenna directly heard that node.

Three different privacy questions​

1. Is the LoRa message encrypted? Meshtastic channel keys affect whether others can read the message. Publicly known keys are not private. Direct-message security follows different mechanisms.

2. Is the MQTT broker connection encrypted? protects the network connection between a client and server when enabled and correctly configured. It does not mean the broker operator cannot see the data it receives.

3. What is published to the broker? The MQTT module can publish encrypted or unencrypted content, and optional JSON output is not encrypted end-to-end. Map-reporting options can disclose a node's identifying details and position with a configurable precision. Do not assume a private channel key automatically prevents other exports from revealing data.

The official MQTT module and Meshtastic encryption documentation explain these separate layers.

Why use MQTT anyway?​

MQTT is genuinely useful for non-emergency integration and for learning distributed networking:

  • Connecting separated radio communities when an Internet-assisted path is acceptable.
  • Sending supported packets to dashboards and monitoring services.
  • Integrating selected environmental measurements with tools such as Home Assistant or Node-RED.
  • Investigating whether a node or gateway is online without making every measurement a direct radio test.

It is also reasonable to choose not to use MQTT. A radio-only mesh is simpler and has fewer Internet-related dependencies. No community should imply that a beginner must join its specific broker in order to use Meshtastic.

A repeatable resilience experiment​

Use two compatible radios that can directly communicate and a gateway you control. Test the following, without disrupting other people's equipment:

  1. Establish successful RF-only messaging with MQTT disabled.
  2. Enable a carefully chosen uplink, confirm packets appear at your own authorized broker or dashboard, and record the results.
  3. If intentionally testing downlink, ensure it is authorized and won't burden a shared channel.
  4. Disconnect only the gateway's Internet access while leaving the radio and local power running.
  5. Send a local RF test. Confirm whether it still arrives.
  6. Check that the MQTT reporting/remote-community path is absent or delayed.
  7. Restore Internet and check recovery—without assuming missing messages will be replayed.

You have now separated radio resilience from online service resilience.

Common questions​

“If I run my own MQTT broker, is it off-grid?” Only if the broker and the IP network needed to reach it can operate independently of the infrastructure you're concerned about. It remains IP-dependent, not a radio-only MeshCore or Meshtastic link.

“Why do I see a node hundreds of miles away?” It may be through MQTT, a gateway, a map or other reporting system. Verify how it was actually heard before calling it an RF contact.

“Does MQTT fix weak antennas?” No. The first and last radio hops still require workable RF propagation unless the relevant devices communicate by some other means.

“Should every node publish GPS?” No. Review position sharing and MQTT map reporting separately in our location/privacy guide.

Source material​

Next: Sensor and environmental telemetry, or test off-grid troubleshooting.

Have a question about your own MQTT gateway or broker? Discuss it in the forum.