[tor-bugs] #18332 [Tor]: Relays should store HS descriptor without the complicated "am I the right one" logic

[Tor Bug Tracker & Wiki]

#18332: Relays should store HS descriptor without the complicated "am I the right
one" logic
-------------------------+------------------------------------
 Reporter:  dgoulet      |          Owner:
     Type:  enhancement  |         Status:  needs_review
 Priority:  Medium       |      Milestone:  Tor: 0.2.8.x-final
Component:  Tor          |        Version:
 Severity:  Normal       |     Resolution:
 Keywords:  tor-hs       |  Actual Points:
Parent ID:               |         Points:  small
 Reviewer:               |        Sponsor:  SponsorR
-------------------------+------------------------------------

Comment (by arma):

 Replying to [comment:6 nickm]:
 > It appears that e167910fce2b83d9de3a252cdf02cabdafced14b implements the
 change, and the other patches on your branch just do cleanups and deadcode
 removals.  What would you say if I took e167910fce in 0.2.8, and the rest
 in 0.2.9?

 Fine by me. The main reason to do it all now is so both branches are
 simpler, in case that becomes useful later (e.g. we have to fix a bug).
 But that's always the reason to do all the changes now now now, and this
 isn't any more compelling a case than the others.

 > Other notes:
 > e167910fce2b83d9de3a252cdf02cabdafced14b rip out
 hid_serv_responsible_for_desc_id()
 >  * NM.1 Does this make bridges start storing hs descriptors?

 I believe it does not change the situation. Bridges used to be willing to
 store hs descriptors (if you give them one whose identifier is
 sufficiently near the bridge's identity key), and they are after this
 change too. Maybe somebody should open a ticket. :)

 The commit *almost* changes things for clients, since it does indeed no
 longer do the {{{if (!server_mode(get_options()))}}} check inside
 router_get_my_routerinfo(). They would need to have an open dirport, and
 you would need to be doing the rendezvous2 post via begin_dir, but
 begin_dir requires an ORPort, so it's not a client.

 > 53902963383e1babfccb8a4ffc7ed4e8accf2214 rip out
 hid_serv_acting_as_directory()
 >  * NM.2 I wonder if this should call public_server_mode() instead of
 calling nothing; I wouldn't want bridges or clients or hidden services to
 start storing descriptors.

 I think this is a fine plan. I'd want to do the check in
 directory_handle_command_post(). And then do the check again in
 rend_cache_store_v2_desc_as_dir() if we like redundancy.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/18332#comment:8>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online