During a review we noticed that the semantics of bootstrap_time in the current openvpn spec might not be exactly compatible with psiphon or Tor.
This needs to be revisited, ideally before next release window.
For consistence, the easy thing to do seems to be to include the total time to bootstrap, and define this field so that it is not included in the case of an error.
However, a valid point still holds about capturing timing of handshake packets in the case of OpenVPN. This can be added as handshake time - the rationale is that further use of the tunnel will add complexity for analysis to extract timing info from the network_events array.
max_runtime could also be added to be sure that we're analyzing equivalent timeout times.
During a review we noticed that the semantics of
bootstrap_timein the currentopenvpnspec might not be exactly compatible with psiphon or Tor.This needs to be revisited, ideally before next release window. For consistence, the easy thing to do seems to be to include the total time to bootstrap, and define this field so that it is not included in the case of an error.
However, a valid point still holds about capturing timing of handshake packets in the case of OpenVPN. This can be added as handshake time - the rationale is that further use of the tunnel will add complexity for analysis to extract timing info from the network_events array.
max_runtime could also be added to be sure that we're analyzing equivalent timeout times.