[tor-bugs] #19522 [Core Tor/Tor]: HS intro circuit retry logic fails when network interface is down

[Tor Bug Tracker & Wiki]

#19522: HS intro circuit retry logic fails when network interface is down
--------------------------+------------------------------
 Reporter:  asn           |          Owner:
     Type:  defect        |         Status:  new
 Priority:  Medium        |      Milestone:  Tor: 0.2.???
Component:  Core Tor/Tor  |        Version:
 Severity:  Normal        |     Resolution:
 Keywords:  tor-hs        |  Actual Points:
Parent ID:  #16387        |         Points:  1.5
 Reviewer:                |        Sponsor:  SponsorR-can
--------------------------+------------------------------

Comment (by timonh):

 Replying to [comment:1 asn]:
 > Suggested fix #1:
 >
 >   In the snippet above, if `rend_service_launch_establish_intro()`
 fails, we '''don't''' remove the intro point from `service->intro_nodes`
 or free it. Instead, we just `continue` to the next intro point. This way
 we will keep on retrying the old intro points every second, till the
 network comes back up.

 Your fix seems to work for me. I switched the network interface and ended
 up with the same intro points. See the attached log.

 > The ''main assumption'' in the above fix is that
 `rend_service_launch_establish_intro()`  can only fail for local reasons
 (bugs, OOM, interface down, etc.). If this is the case, then there is no
 reason to blame the intro point for this failure and we can just keep on
 retrying it till the local issues get resolved.
 >
 > We should verify the assumption above before implementing the suggested
 fix.

 If the assumption can be verified it would be nice to have the fix soon.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/19522#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online