[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-bugs] #27416 [Core Tor/DocTor]: Automatc tickets for malformed descriptors (was: Improve descriptor validation in Tor using Stem)

#27416: Automatc tickets for malformed descriptors
-----------------------------+----------------------------------
 Reporter:  teor             |          Owner:  atagar
     Type:  defect           |         Status:  new
 Priority:  Medium           |      Milestone:  Tor: unspecified
Component:  Core Tor/DocTor  |        Version:
 Severity:  Normal           |     Resolution:
 Keywords:  tor-dirauth      |  Actual Points:
Parent ID:                   |         Points:
 Reviewer:                   |        Sponsor:
-----------------------------+----------------------------------

Comment (by atagar):

 > Can we run a scheduled task in Jenkins or Travis?

 Hi teor, interesting suggestions. Jenkins/Travis would be great fits for
 Stem integ tests to catch tor regressions but I'm unsure it would be a
 good fit for malformed descriptor notices.

 Jenkins/Travis are great for 'x regressed, and will remain broken until
 action is taken', but malformed descriptors tend to be transient. Unless
 I'm missing something we need an asynchronous notification mechanism like
 a ticket to capture those.

 I've conflated two similar but fundamentally different issues in this
 ticket. Sorry about that! Gave this more thought and think I have a decent
 path forward on both the issues discussed here...

 1. **Malformed descriptors** should result in a ticket to the network
 team. To streamline this I'll adjust our DocTor descriptor validity check
 to automatically file tickets rather than email me.

  These tickets should be rate limited to 1/day (to avoid potential messes,
 such as from a botched sybil attack), attach the bad descriptor, and
 provide an explanation of why it's malformed.

  This ticket will track this, and the **next step is in my court**.

 2. **Tor regressions** should be addressed by proactively running our
 tests prior to pushing. We already have a
 [https://jenkins.torproject.org/job/stem-tor-ci/ Jenkins job] that tests
 after each tor and stem push, but if the network team would care to adjust
 it or integrate with Travis I'm game.

  I've asked Nick and David to test prior to pushing in the future so for
 me the next step on this front is to see if I stop getting notified of tor
 regressions. If so, great. If not then I'll file a **separate** ticket to
 discuss improving our workflow.

 > We only recently set up the team rotations. We're going to talk about
 them
 > again in Mexico. Let's make sure we put stem and consensus-health on the
 > list.

 Gotcha! Hat's off to the network team by the way. I won't be in Mexico,
 but you guys are doing a phenomenal job at integrating bests practices and
 improving our processes.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/27416#comment:5>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online

_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs