[tor] noisetor administrivia

[Andy Isaacson]

We discussed a few issues:

1. how to respond to abuse complaints
 - we have template messages,
   https://github.com/radii/noisetor-message-templates
 - if a complaint was written by a person, please write them a personal
   response (and potentially also include an appropriate template)
 - every message should get a response
 - if Quadranet's ticketing system is involved, be careful to put the
   ticket number in the Subject (or the response will cause another
   ticket).
2. monitoring
 - we have munin running on both noisetor and exit01
 - use rsync to copy to localhost for viewing
   rsync -av noisetor01:/var/cache/munin/www/ /tmp/www/
 - logwatch running on both noisetor and exit01
 - apticron
3. known issues
 - our tor instances occasionally grow several hundred MB and cause a
   ton of swap.  All 4 instances go at about the same time, so it seems
   like it's some network-triggered event.  It's unclear what triggers
   this, but it seems to get better after restarting the tor processes.
   We saw three such spikes this week, which is much more frequent than
   normal.
 - our throughput varies a fair amount, even when not seeing the dips
   caused by the "swap event".  We routinely peak at 470 Mbps and trough
   at 350 Mbps with a very regular 24 hour pattern to the graph.
4. housekeeping
 - keep packages up to date with apticron + 'apt-get dist-upgrade'
 - respond to abuse reports within 48 hours
 - watch donations graph and summary, ensure we're not going negative
 - watch munin graphs, ensure we're still pushing bits and not
   crashing/swapping
5. how to update the webpages with git
 setup:
 - git clone /var/git/www.git public_html
 to edit and publish:
 - cd public_html
 - vi index.html
 - git commit -am 'updated bandwidth summary numbers'
 - make install
    * this pushes your changes to /var/git/www.git
    * a post-receive hook in that repo pushes the bits to /var/www

-andy