Skip to content

fix(reticulum): exhaust alternate paths for Nomad and LXMF Direct #766

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
rinchen merged 2 commits into from
Aug 2, 2026
Merged

fix(reticulum): exhaust alternate paths for Nomad and LXMF Direct #766

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5038c37
fix(reticulum): exhaust alternate paths for Nomad and LXMF Direct
rinchen Aug 2, 2026
222701a
fix(reticulum): tighten multipath failover review feedback
rinchen Aug 2, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions docs/troubleshooting.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1034,7 +1034,7 @@ In dev, **Start stack** now rebuilds when `reticulum-sidecar/src/**/*.rs` or `Ca

Unrecognized codes pass through unchanged.

TCP/network Nomad Links use path-scaled initiator hops (`link_hops = clamp(path_hops, 3, 7)`) and a LinkClient proof wait of the **remaining overall MeshChat deadline** (~45s TCP after instant pubkey recall), matching v5.25.0. Do not cap LRPROOF at hops×6 or a 30s floor — that false-failed multi-hop hub pages that still load on release. First attempts use a cached path when present (no DropPath storm); missing paths RequestPath briefly and may return `path_timeout`. Retries may DropPath + rediscover; `force_path_ok=true` means rediscovered after absence only (cache hits log `force_path_ok=false`). Failure logs (`[nomadNetworkStore] … fetch failed` and sidecar `Nomad Link query failed`) include `path_hops`, `link_hops`, `proof_budget_secs`, `force_path_ok`, `path_ensure_kind`, `elapsed_ms`, and `raw=`. UI errors distinguish cached-path vs rediscovered-path link failures.
TCP/network Nomad Links use path-scaled initiator hops (`link_hops = clamp(path_hops, 3, 7)`) and a LinkClient proof wait of the **remaining overall MeshChat deadline** (~45s TCP after instant pubkey recall), matching v5.25.0. Do not cap LRPROOF at hops×6 or a 30s floor — that false-failed multi-hop hub pages that still load on release. First attempts use a cached path when present (no DropPath storm); missing paths RequestPath briefly and may return `path_timeout`. On TCP `link_timeout`, the sidecar suppresses the dead iface, drops the failed via, promotes ranked path-slot backups / other live hubs (extra RequestPath when another TCP/RF iface is up), then retries inside the same fetch. LXMF Direct chat uses the same exhaustion before the one-shot preferred-PN fallback. `force_path_ok=true` means rediscovered after absence only (cache hits log `force_path_ok=false`). Failure logs (`[nomadNetworkStore] … fetch failed` and sidecar `Nomad Link query failed`) include `path_hops`, `link_hops`, `proof_budget_secs`, `force_path_ok`, `path_ensure_kind`, `elapsed_ms`, `tried_interfaces`, `failover_rounds`, `iface`, and `raw=`. UI errors distinguish cached-path vs rediscovered-path link failures.

**Cause**: Older `LinkClient` always waited for a fresh path-response announce for the destination public key, even when Nomad announces had already cached it. Successful fetches could also deregister all `nomadnetwork.node` announce handlers. Distant/high-hop nodes can still time out at the path stage (expected RF/mesh reachability limits).

Expand All @@ -1044,7 +1044,7 @@ TCP/network Nomad Links use path-scaled initiator hops (`link_hops = clamp(path_
2. Rebuild sidecar: `pnpm run reticulum:sidecar:build`, restart stack.
3. Prefer low-hop nodes while testing; hop count is shown in the Nomad list.
4. Match the humanized message to the table above — `path_timeout` / high hops often mean RF reachability limits, not a mesh-client bug.
5. For TCP `link_timeout`, check log fields `path_hops` / `link_hops` / `proof_budget_secs` / `raw=` — UI hop counts can lag the path table; trust `path_hops`. Persistent fails after the full proof budget usually mean the peer/hub did not return LRPROOF.
5. For TCP `link_timeout`, check log fields `tried_interfaces` / `failover_rounds` / `iface` first (primary signal after path failover), then `path_hops` / `link_hops` / `proof_budget_secs` / `raw=` — UI hop counts can lag the path table; trust `path_hops`. Persistent fails after the full proof budget usually mean the peer/hub did not return LRPROOF.

### Reticulum sidecar stops during dev (Vite HMR)

Expand Down
Loading