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

Re: [tor-bugs] #29787 [Metrics/Onionperf]: Enumerate possible failure cases and include failure information in .tpf output



#29787: Enumerate possible failure cases and include failure information in .tpf
output
-------------------------------+------------------------------
 Reporter:  karsten            |          Owner:  metrics-team
     Type:  enhancement        |         Status:  new
 Priority:  Medium             |      Milestone:
Component:  Metrics/Onionperf  |        Version:
 Severity:  Normal             |     Resolution:
 Keywords:                     |  Actual Points:
Parent ID:                     |         Points:
 Reviewer:                     |        Sponsor:
-------------------------------+------------------------------

Comment (by karsten):

 Replying to [comment:24 acute]:
 > Replying to [comment:23 karsten]:
 > > I made a hopefully even more useful plot that distinguishes between
 public and onion server and that groups errors by OnionPerf/tgen (prefix:
 TGEN) and tor (prefix: TOR). Admitted, it's a lot of information in a
 single plot. It probably needs a fair amount of textual explanation to be
 readable for Tor Metrics users.
 > > However, I'm afraid that I'm running out of time for the failure
 analysis.
 >
 >
 > I'm happy to come up with a couple of paragraphs to describe each
 failure case - already have explanations for some of the proxy errors.
 Would that be helpful?

 Yes, that would be very helpful!

 Regarding lengths, I think that we could use different lengths depending
 on intended audience:
  - very short, user-oriented explanations for the graph page,
  - medium-sized, one or two sentence explanations for the .tpf format
 specification page, and
  - as-detailed-as-we-want explanations on Trac tickets where we discuss
 failure cases and why they occurred.

 So, yes, anything you write about these cases will be useful!

 > > I'm wondering how we could add a graph like this to Tor Metrics. We
 earlier speculated about including more failure information in .tpf files,
 but we would then have to produce new .tpf files for past measurements. Is
 that trivial to do? Or would we have to switch to .json files for Tor
 Metrics?
 >
 > The tpf files are built from the json files at the time of analysis.
 This means we have to change the analysis code and then rerun the analysis
 for every single pair of logs. Do we have all the logs?
 > Current PRs to be merged into the analysis code relate to #29374, and
 immediately after that #26673, then this can be my next priority for the
 analysis.

 Sounds perfect! Yes, I think we still have all the logs, and I can
 reprocess them here. I'll wait for the changed analysis code then. Thanks!

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/29787#comment:25>
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