[metrics-bugs] #30217 [Metrics/CollecTor]: Extend CollecTor filesystem protocol for bandwidth files

[Tor Bug Tracker & Wiki]

#30217: Extend CollecTor filesystem protocol for bandwidth files
-------------------------------------------------+-------------------------
 Reporter:  irl                                  |          Owner:
                                                 |  metrics-team
     Type:  enhancement                          |         Status:  new
 Priority:  Medium                               |      Milestone:
Component:  Metrics/CollecTor                    |        Version:
 Severity:  Normal                               |     Resolution:
 Keywords:  tor-bwauth,tor-dirauth,metrics-      |  Actual Points:
  roadmap-2019-q2                                |
Parent ID:  #21378                               |         Points:
 Reviewer:                                       |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by karsten):

 I didn't give this as many thoughts as you all did, but as I see it, we
 would want to include the following information in the bandwidth file name
 as archived by CollecTor:
  1. '''Timestamp:''' The timestamp (first line in the document) is not
 guaranteed to be unique between bandwidth generators and apparently not
 even for the same bandwidth generator. Still, including the timestamp is
 useful and also required for things like monthly tarballs.
  2. '''Identifier:''' Having an identifier of the bandwidth generator
 would certainly be useful. But it sounds like we don't have one for
 existing bandwidth files, which we want to archive anyway. I think we
 might have to find a bandwidth file name that doesn't include a bandwidth
 generator identifier, and whoever wants to only look at bandwidth files
 produced by certain bandwidth generators will have to use `grep`.
  3. '''Digest:''' If we include a digest, we'll come up with a unique file
 name, which is what we want. Ideally, we'd use the same digest that is
 also included in votes, though we might have to encode it as hex rather
 than base64 to avoid all kinds of trouble with `/` and `+` in file names
 and URLs and upper- and lower-case in case-insensitive file systems.

 Putting everything together, how about this: `2019-04-21-09-00-00
 -bandwidth-file-
 42D07217935283D42CC559A61E7F788E3A92F0E2CB54BE58B5B83E25563C55A2` ?

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/30217#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online