[tor-bugs] #31385 [Circumvention/Snowflake]: Snowflake client fails after bootstrap

[Tor Bug Tracker & Wiki]

#31385: Snowflake client fails after bootstrap
-------------------------------------+--------------------------
 Reporter:  cypherpunks              |          Owner:  cohosh
     Type:  defect                   |         Status:  assigned
 Priority:  Medium                   |      Milestone:
Component:  Circumvention/Snowflake  |        Version:
 Severity:  Normal                   |     Resolution:
 Keywords:                           |  Actual Points:
Parent ID:                           |         Points:
 Reviewer:                           |        Sponsor:
-------------------------------------+--------------------------

Comment (by cohosh):

 Replying to [comment:14 arma]:
 > Replying to [comment:12 cohosh]:
 > > if the client has already sent its upstream bytes, those bytes are
 lost forever and so all subsequent snowflakes will look like they are
 failing even if their connection to the bridge is fine. This is what the
 sequencing layer in #29206 is for.
 >
 > Would it be smart, while we don't have the sequencing layer in place
 yet, for Snowflake to have some keepalive or timeout feature, where it
 notices that it's sent its bytes, and things aren't looking good, and it
 should abort that connection so Tor can try a new one?
 >
 > If we can do it in a simple way, it might help people in the short term.
 And in the long term, we might still need some sort of similar "gosh that
 channel looks like it has failed" feature to know when it's time to launch
 a new one.
 I'm working on #29206 this week. I'll keep this in mind, but the answer
 might be that it's just as fast to go ahead and implement the sequencing
 layer (which is already a very simple version of what we'll actually want
 in the future).

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/31385#comment:16>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online