This is a work in progress and is partially available. Stability is not guaranteed.
:::
The Store Forward Module is an implementation of a Store and Forward system to enable resilient messaging in the event that a client device is disconnected from the main network.
Because of the increased network traffic for this overhead, it's not advised to use this if you are duty cycle limited for your airtime usage nor is it advised to use this for SF12 (Long Range / Slow).
To use / test this you will want at least 3 devices
- One device will (currently) need be a tbeam v1.0 and tbeamv1.1 configured as a Meshtastic router. Other devices with built in PSRAM will be supported at some point.
- Two others will be regular clients. Nothing special required.
Don't use this on the "Long Range / Slow" or "Long Range / Fast" channel settings. You're welcome to try and report back, but those channels have a [very low bitrate](/docs/developers/firmware/radio-settings#pre-defined).
-`store_forward_module_enabled` - Set this to true to enable the module. False to disable the module.
- Optional configuration
-`store_forward_module_records` - Set this to the maximum number of records to save. Best to leave this at the default (0) where the module will use 2/3 of your device's available PSRAM. This is about 11,000 records.
- Name your router node something that makes it easily identifiable, aka "Router".
Don't enable the Store and Forward module on multiple routers!
### Client Usage
Currently, no special configuration is required. To request your history sent to you, send the command into the message field "SF". That's it. This will eventually change to make it easier. At the moment, that message will be sent to everyone on the mesh but we'll (eventually) make it easier to use where there'll be a button (or maybe it'll be transparent) and the command isn't sent as a text message to the mesh.
| SF | Send the last few messages I may have missed |
| SFm | Send a 240 byte payload (Used for testing) |
The Store and Forward module will only service one client at a time. If a second client requests messages while the S&F is busy, the S&F will send a private message to the second client that they will need to wait.
Story: Carol has been away from the mesh with device turned off. She would like to get a replay of what she has missed. The router will return the messages.
Story: Carol has been away from the mesh with device turned off. She would like to get a replay of what she has missed but the router indicates there are no messages available.
Story: The Store & Forward Router sends a periodic message onto the network. This allows connected devices to know that a router is in range and listening to received messages. Client will not respond to network but (optionally) indicate to the user that a S&F router is available or not available.
- StoreAndForward.heartbeat.Period = 120 // Expect a heartbeat every 2 minutes.
- StoreAndForward.heartbeat.Secondary = false // If true, this is a secondary "backup" S&F node. Will be (eventually) used for router election in the event there are multiple S&F Routers.
When a message is 'real time', time a message is sent is the same as the time the message was received. In the case the message is delayed, there is no timestamp available for the message.
Where the message is a delayed broadcast, the "To" is _not_ a broadcast address but rather the address of the device that requested the messages to be replayed. This is to allow the message to be routed over the mesh network and not displayed in the message screen of other devices.
## Developer TODO
Not necessarily in this order:
- Client Interface (Web, Android, Python or iOS when that happens) to request packets be resent.
- Router sends a heartbeat so the client knows there is a router in range.
- Support for a mesh to have multiple routers that have the store & forward functionality (for redundancy).
- Add a default channel at about 1.5kbit / sec.
- Eventually we could add a "want_store_and_forward" bit to MeshPacket and that could be nicer than whitelists in this module. Initially we'd only set that bit in text messages (and any other module messages that can cope with this). This change would be backward wire compatible so can add easily later.
- Have a "cool down" period to disallow a client to request the history too often.
- Message with SF status on requests.
- Be able to identify if you're within range of a router.
- Be able to specify the HOP MAX to reduce airtime. Is this necessary?
- Restrict operation of S&F on the slow channel configurations.
- Create a TX queue to prevent the history from being modified in flight.
- Only allow n-number of requests to the router at any one time.
- Set number of max messages from the history.
- Calculate a new channel with 250kHz bandwidth and ~1.5kbit.
\*\*\* Done
- Disable ACK. If the router is within range, so is the requester.
- Currently the way we allocate messages in the device code is super inefficient. It always allocates the worst case message size. Really we should dynamically allocate just the # of bytes we need. This would allow many more MeshPackets to be kept in RAM.
- Allow max history to be defined by radioConfig.preferences.store_forward_module_records.
- Add a starting to send / finished sending message.