In PR #229 review, we noted a follow-up: goroutines in StartDKG / StartReshareDKG* wait on p.WaitEnd() and can block forever if the DKG protocol stalls (lost messages / deadlock).
This is not HTTP-handler blocking (already addressed via timeouts), but it is a resource leak under the same stall/DoS model.
Ideas to investigate:
- Add a timeout around waiting for
WaitEnd() and propagate an error to initiator.
- Check whether the underlying kyber protocol supports cancellation/abort; if so, wire cancellation to instance lifetime.
Context: PR #229 (F-ssv-dkg-007).
In PR #229 review, we noted a follow-up: goroutines in
StartDKG/StartReshareDKG*wait onp.WaitEnd()and can block forever if the DKG protocol stalls (lost messages / deadlock).This is not HTTP-handler blocking (already addressed via timeouts), but it is a resource leak under the same stall/DoS model.
Ideas to investigate:
WaitEnd()and propagate an error to initiator.Context: PR #229 (F-ssv-dkg-007).