You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Those tests started randomly failing after moving to anchor outputs.
The reason is that anchor outputs have a bigger commit weight than non
anchor channels, and with our initial balance allocation we sometimes
reached a point where the *remote* had exhausted all of its balance to
pay commit fees (when the feerate was around 10 000 sat/kw), so the
*local* couldn't send HTLCs (`RemoteCannotAffordFeesForNewHtlc`). This
is an expected situation where we cannot do anything except wait for
pending HTLCs to resolve before sending other HTLCs.
We simply increase the initial balance to avoid running into this case.
We also remove the `skip` case (which wasn't working at all, we should
have used the `cancel` scalatest helper if we wanted to interrupt the
test without failing it) and allow failures during the initial HTLC
setup (which can happen if HTLCs are larger than the available balance).
0 commit comments