Skip to content

Halo direct path retries - #1

Merged
usrflo merged 14 commits into
usrflo:halo-direct-path-retriesfrom
mikecarper:halo-direct-path-retries
Apr 30, 2026
Merged

usrflo merged 14 commits into
usrflo:halo-direct-path-retriesfrom
mikecarper:halo-direct-path-retries

Conversation

@usrflo

@usrflo usrflo commented Apr 30, 2026

Copy link
Copy Markdown
Owner

source: mikecarper/MeshCore:halo-direct-path-retries

@usrflo
usrflo merged commit 074cba3 into usrflo:halo-direct-path-retries Apr 30, 2026
usrflo added a commit that referenced this pull request Sep 6, 2026
Field feedback (2 testers, days of operation): ~96% of coverage TRACEs
never return (sent=1443 ret=46 tmo=691 neg=706) while adverts/floods on
the same links flow fine. Root cause #1: trace_tx_power_dbm defaulted to
10 dBm, so hop-1 M->a of every probe ran ~12 dB below the tx_power_dbm
links the coverage graph is supposed to represent -- a bidirectional
flood link systematically fails the probe, poisoning (a,b) as no-edge
for ~10h and over-excluding reach (unr).

Default trace power now follows the node's tx_power_dbm (the lower/
restore burst window goes dormant; `set trace.tx.power` remains for
operators who prefer less disturbance over fidelity). Note in comments:
setTxPower is radio-global (floods queued inside the 2s burst window
were weakened too), and a >=3s retry fired after the 2s restore, i.e.
at a different power than its initial attempt.

Sim is power-neutral (SimRadio ignores TX power), so validation is the
unchanged clique smoke (sent=3 ret=2 edge=2, identical to baseline);
field units re-verify via `near` (ret/edge should rise, unr fall).
Units that ever ran a `set` command keep the persisted 10 dBm.

Co-Authored-By: Claude <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants