1. 19 Aug, 2013 4 commits
  2. 18 Aug, 2013 6 commits
  3. 17 Aug, 2013 3 commits
  4. 16 Aug, 2013 1 commit
  5. 15 Aug, 2013 2 commits
  6. 11 Aug, 2013 1 commit
    • Damian Johnson's avatar
      Start of a script to perform DocTor checks · 1e49c330
      Damian Johnson authored
      First step toward replcing DocTor...
      
      https://lists.torproject.org/pipermail/tor-dev/2013-August/005233.html
      
      This performs the following checks from its Warning.java...
      
        * NoConsensusKnown
        * ConsensusDownloadTimeout
        * ConsensusNotFresh
        * ConsensusMethodNotSupported
        * DifferentRecommendedClientVersions
        * DifferentRecommendedServerVersions
        * UnknownConsensusParams
      
      Remaining checks left to be done are...
      
        * ConflictingConsensusParams
        * CertificateExpiresInThreeMonths
        * CertificateExpiresInTwoMonths
        * CertificateExpiresInTwoWeeks
        * VotesMissing
        * BandwidthScannerResultsMissing
        * ConsensusMissingVotes
        * ConsensusMissingSignatures
        * MissingAuthorities
        * UnexpectedFingerprints
        * UnrecommendedVersions
      1e49c330
  7. 04 Aug, 2013 3 commits
  8. 30 Jul, 2013 7 commits
    • Damian Johnson's avatar
      Function to log stem debug logs · 9f6fa3da
      Damian Johnson authored
      We're getting a stacktrace that indicates a stem bug...
      
      Traceback (most recent call last):
        File "/home/atagar/Desktop/tor/tor-utils/descriptor_checker.py", line 90, in <module>
          main()
        File "/home/atagar/Desktop/tor/tor-utils/descriptor_checker.py", line 52, in main
          log.debug("  %i descriptors retrieved from %s in %0.2fs" % (count, query.download_url, query.runtime))
      TypeError: float argument required, not NoneType
      
      We'll frequently be wanting stem debug logs so adding a function to get them.
      Actually, we kinda lucked out since my first run after adding this experienced
      the issue. Here's the corresponding stem logs...
      
      07/30/2013 10:14:25 [TRACE] Descriptors retrieved from 'http://171.25.193.9:443/tor/server/all.z' in 19.21s
      07/30/2013 10:15:07 [DEBUG] Unable to download descriptors from 'http://154.35.32.5:80/tor/extra/all.z' (2 retries remaining): Error -5 while decompressing data: incomplete or truncated stream
      07/30/2013 10:15:40 [DEBUG] Unable to download descriptors from 'http://212.112.245.170:80/tor/extra/all.z' (1 retries remaining): Error -5 while decompressing data: incomplete or truncated stream
      07/30/2013 10:16:13 [DEBUG] Unable to download descriptors from 'http://212.112.245.170:80/tor/extra/all.z': Error -5 while decompressing data: incomplete or truncated stream
      
      So... guess the compression issue isn't as solved as I thought. This isn't
      gonna be fun to troubleshoot...
      9f6fa3da
    • Damian Johnson's avatar
      Checking for email password relative of scripts too · 39223e06
      Damian Johnson authored
      Oops, missed one of our resources. Fetching this relative of our scripts rather than the execution location too.
      39223e06
    • Damian Johnson's avatar
      Logging traceback on error · ed0373ec
      Damian Johnson authored
      We're getting some silent failures in the logs. Printing the stacktrace to help
      troubleshoot these.
      ed0373ec
    • Damian Johnson's avatar
      Minor pydoc corrections · 7cf79534
      Damian Johnson authored
      7cf79534
    • Damian Johnson's avatar
      Sorting relays by nickname · 7fc99ee6
      Damian Johnson authored
      Having the sybil checker sort the list of relays it sends by nickname. Also
      tweaking the layout a bit to be more condensed.
      7fc99ee6
    • Damian Johnson's avatar
      Placing resources relative of scripts · e84e2770
      Damian Johnson authored
      Oops, meant for the logs and data directories to be relative of these scripts,
      not where they were executed from.
      e84e2770
    • Damian Johnson's avatar
      Replacing consensusTracker.py with a simple sybil checker · 9b7de30f
      Damian Johnson authored
      Our consensusTracker.py was petty slick, but most of what it does is
      unnecessary. The html formatted emails were pretty, and the stats were kinda
      interesting. But the only true usefulness of the script was to notify us of
      sybil attackers spamming us with relays. Replacing it with a far smaller,
      simpler script that just does that.
      9b7de30f
  9. 29 Jul, 2013 4 commits
  10. 28 Jul, 2013 2 commits
  11. 04 Feb, 2013 2 commits
  12. 12 Dec, 2012 3 commits
  13. 07 Dec, 2012 1 commit
    • Damian Johnson's avatar
      Migrating the consensus tracker from TorCtl to Stem · 2e42c5dd
      Damian Johnson authored
      The consensus tracker was a hack job. It does the trick in terms of notifying
      us of basic Sybil attacks on the network, but it's only meant as a short term
      solution until we invest time into a proper monitor and alarm framework.
      
      Migrating the script from TorCtl to Stem. This is a pretty direct migration -
      the behavior is unchanged and I'm not trying to clean up the script (though
      stem does make it a bit more readable).
      
      Doing this migration revealed a catastrophic performance bug in stem that is
      now fixed...
      
      https://gitweb.torproject.org/stem.git/commitdiff/bbd702c81e1923bfbab236d0d3649fd2f87b3d95
      
      The memory usage is still a fair bit higher (around 5.1% of the memory on my
      netbook verses 1.9% before). However, this isn't entirely unexpected. In a lot
      of cases TorCtl really didn't do much, providing us raw strings rather than
      parsed objects. Still, there's certainly plenty of low hanging fruit we should
      address in stem to lower that gap.
      2e42c5dd
  14. 05 Dec, 2012 1 commit