#24716: Try cranking up cbttestfreq consensus param, to see if it helps the
current
overload
------------------------------+------------------------
Reporter: arma | Owner: (none)
Type: task | Status: new
Priority: Medium | Milestone:
Component: Core Tor/DirAuth | Version:
Severity: Normal | Resolution:
Keywords: | Actual Points:
Parent ID: | Points:
Reviewer: | Sponsor:
------------------------------+------------------------
Comment (by arma):
I also just added {{{cbtmintimeout=5000}}} to moria1's consensus params,
on the theory that if clients really were learning super low timeouts and
they weren't unlearning them, then we'll make them more likely to be
willing to use the circuits they make.
This choice is not uniformly a win, since with this raised min, clients
that had computed a super-fast cbt will now be less likely to abandon it
(and move back to the slow learning phase, during which they use a timeout
of 60s). But it could help us in the hypothesis 2 case, where due to bugs
or luck clients are sticking to their super-fast cbt despite 80% circuit
failure.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/24716#comment:1>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
[email protected]
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs
