MeshMonitor v4.17.0 is out after nine release candidates. It is the largest release since 4.0: more than 60 features and 60 fixes. The themes are seeing one mesh through several sources at once, far more from MeshCore repeaters, and automations that know what time it is.

Highlights below. The full list is in the CHANGELOG. The Docker image also changed in ways that matter to anyone who runs it; that, and a round of permission fixes, has its own post.

One mesh, many sources

MeshMonitor now works out when two of your sources heard the same packet, and uses that everywhere.

• "Heard here" map edges draw a link when one source's radio is heard by another, and mark links a traceroute has confirmed in both directions. You see an edge only when you can read both sources.
• Unified Messages has an All Channels view with a channel tag on each message, merges one MeshCore message heard by several sources into one row, and colours each message's left border by mesh: Meshtastic, MeshCore or Reticulum. The same stripe marks messages in each source's own channel and DM views.
• One notification per packet. Several sources hearing the same message no longer means several alerts.
• Packet Monitor and Coverage tag a packet that one of your sources sent, or likely relayed, and another heard.

MeshCore

• Repeater sources show much more. A repeater on a USB cable now reports its own telemetry and neighbours on stock firmware. With a firmware build that has packet logging on, it also fills Nodes from adverts, Packet Monitor, and channel Messages for channels whose key MeshMonitor holds.
• Trace SNR history. Every TRACE a Companion radio hears is stored hop by hop, sent by you or not. A Repeater source does the same with a packet-logging build. A contact's details show each link direction as a summary row and as a full chart over 24 hours to 90 days. Automations can trace paths on a schedule to keep the history fresh.
• Mentions. A new notification fires when a channel message or room post names your node, at most once per sender per source every five minutes. It is off until you turn it on.
• Channel database. MeshCore channel secrets can live server-side, so MeshMonitor decrypts group traffic in the Packet Decode view and on repeater and MQTT ingest sources. A new entry starts with no user access until you grant it.
• Also: resend for channel messages no repeater echoed, real delivery state on your own DMs and room posts, an auto return path for Trace Path, Discover Path feedback, favorites from Node Details, long Auto-Acknowledge replies split or cut, and a repeater glyph on the map.

Automations

• Timed events. A scheduled automation can carry a start time, an end time and a time zone. It stays quiet before the start and turns itself off at the end. Scheduled sends now run no more often than every five minutes.
• New triggers and tokens: node discovered and node updated for both protocols, with role, hops, last-heard and age tokens on MeshCore nodes.
• Run-scoped step outputs let one step use what an earlier step produced.
• Forwarding gains a per-source master switch, a sidebar pill that shows its state, and an automation action.
• Auto-Acknowledge on a channel now waits 5 to 30 seconds and stays quiet once two other nodes have answered.

Messaging and notifications

• Quick status. A pill in the header sets your node's Status Message from presets. MeshMonitor now asks the node to confirm it took the status before it reports a save.
• Sender popups. Click a sender's name in chat for the node popup, with a status badge on the avatar.
• Notification format is yours to edit, and the default is lighter.
• Notification settings no longer cross sources. A source you never configured used another source's settings. It now uses quiet defaults. If you run several sources, check each one's Notifications tab after the upgrade.
• Inline message translation, with a cache: a phrase is translated once and shown to every viewer who reads that language. Thanks to Crim for the feature.
• CSV export of filtered messages.

Security and keys

• Reliable PKI primes NodeInfo before a PKI send to a node that has stopped answering. It is off by default.
• Avoid PKI sends telemetry, LocalStats and neighbor requests channel-encrypted. DMs are untouched. It is off by default.
• Proven receipt. With firmware 2.8.1 or later on your own radio, Delivery Details shows whether a DM's ack carries a valid proof from the recipient's key.
• Protection Level picker for the firmware's packet signature policy (firmware 2.8).
• Shutdown, DFU and factory reset in Admin Commands, each written to the audit log. DFU and the resets work on the local node only.
• Per-source permissions are enforced on every route, and a system backup download is now admin only. See the security post.

Maps, reports and the rest

• Local map markers, per source, that are never transmitted, and on-air waypoints straight from Map Analysis.
• Aircraft: mark a node as aircraft or not by hand, and optionally ask a newly flagged aircraft for its position.
• Traceroute Explorer report and a Live Mesh Activity dashboard widget.
• MQTT: a broker preset chooser with the official Meshtastic broker, and finer hop-raise controls per bridge.
• TAK team and role settings, a remote-admin shortcut in node details, and number fields you can clear.
• Prometheus: each node's last position time on /api/v1/metrics. Thanks to @zaepho.

Upgrading

• Migrations 183 to 202 run on first start. Back up first.
• Docker: every process in the container now runs as node, and the image is a third smaller. Most installs need no change. Read the security post if you use cap_drop, a user: line, serial devices or Kubernetes.
• Non-admin users may lose access. A user now needs the grant on the source a request acts on. Grant it on that source under Users if a page goes missing.
• Backups now hold keys and password hashes. Only an admin can download one.
• Notifications: a source you never saved settings for notifies for direct messages only, with Apprise off until you add URLs for it. New-node and traceroute alerts default to off.
• Scheduled automations faster than every five minutes keep running, but must be slowed before they can be saved again. Scheduled MeshCore path traces are exempt and may run each minute.
• Auto-Acknowledge on a channel now answers 5 to 30 seconds late, and not at all once two other nodes have answered.
• API clients: an unknown sourceId returns 404, and fromNodeNum in a traceroute response is always the requester.
• Number fields: blank no longer means 0. Type 0 where 0 means off.

The CHANGELOG has the full upgrade notes.

Thanks to everyone who tested nine release candidates, and to @Cryptoteep for the All Channels view, @Crim for message translation and @zaepho for the metrics work.