ScotMesh RNS Network
Known bugs and workarounds
From the ScotMesh wiki
Real faults that cost real time, with the fix. Add to this whenever something takes more than an hour to work out — these two already have.
i2pd crashes repeatedly with SAM in use
Symptom: i2pd dies every few minutes with SIGSEGV, or malloc(): corrupted top size. Reticulum’s I2P interface then reports Connect call failed ('127.0.0.1', 7656) because the SAM bridge keeps vanishing.
Cause: two use-after-free bugs on I2P destination teardown, triggered by the SAM session churn Reticulum causes when it retries tunnel setup. Present in every released version, including the Ubuntu package and upstream 2.61.0.
Fix: build 2.61.0 with upstream PRs #2497 and #2542 cherry-picked. Two files, eighteen lines. Do not build the branch tip — it is hundreds of commits ahead of the last release.
Clients locked out for 12 hours after a node restart
Symptom: a client sits on “securing link” forever. The node logs nothing at all — not even a failed attempt. Other hubs work fine from the same client.
Cause: BackboneInterface blocks a peer for 12 hours after five connections shorter than 20 seconds. Restarting your node makes every client reconnect at once, which trips it. Look for Ignoring further connections from <ip> due to fast-flapping in the daemon log.
Fix: block_fast_flapping = No on the listener. The default is aimed at pathological backbone peers and is far too harsh for mobile clients, which roam and reconnect normally.
This is a copy of wiki.scotmesh.net/wiki/Known_bugs_and_workarounds, last edited 2026-09-11. To change it, edit the page on the web wiki; this copy follows within 10 minutes.