[tor-bugs] #20082 [Core Tor/Tor]: Lower initial descriptor upload delay for ephemeral services

Tor Bug Tracker & Wiki blackhole at torproject.org
Wed Sep 7 14:26:01 UTC 2016


#20082: Lower initial descriptor upload delay for ephemeral services
------------------------------------------------+--------------------------
 Reporter:  twim                                |          Owner:
     Type:  enhancement                         |         Status:  new
 Priority:  Medium                              |      Milestone:  Tor:
                                                |  0.2.???
Component:  Core Tor/Tor                        |        Version:
 Severity:  Normal                              |     Resolution:
 Keywords:  tor-hs, research, proposal-needed?  |  Actual Points:
Parent ID:                                      |         Points:
 Reviewer:                                      |        Sponsor:
                                                |  SponsorR-can
------------------------------------------------+--------------------------

Comment (by dgoulet):

 Replying to [comment:10 teor]:
 > However, the threat here is that hidden services that have unstable
 introduction points now upload their descriptors 6x more often.
 >
 > Why don't we make the initial upload 5s, and every upload after that
 30s?
 > Or even better, some kind of exponential backoff to a few minutes - if
 you've changed your intro points ten times, we really don't want your
 eleventh descriptor any time soon.

 There is a retry timeout for IP circuits if too many fails (see
 `INTRO_CIRC_RETRY_PERIOD`). We rely on that for an upper limit of
 descriptor upload. If an IP keeps failing after a short period (5
 minutes), then the IP circuit building retry timeout mechanism kicks in
 and thus you won't see a zillion descriptor publication. Maybe that's not
 perfect but that's imo something different from the 30 seconds delay added
 at startup time.

 Now, if your IPs keep failing *after* the 5 minutes retry window (ex:
 circuit is closed because bad network), well you indeed need to rebuild a
 new descriptor with a new IP and publish it but that's OK imo. Adding a
 delay to publication won't help here because we already have that 5
 minutes "wait period" in the first place to avoid too many tries.

 I have this feeling that we might be at the point of going on tor-dev@
 with this discussion because some of us wants an initial delay or get rid
 of it or only do something in 224?

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/20082#comment:12>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list