[tor-dev] strange ARM results

[Rob van der Hoeven]

On Mon, 2016-10-17 at 22:30 +1100, teor wrote:
> > On 17 Oct 2016, at 22:04, Rob van der Hoeven <robvanderhoeven at ziggo.nl> wrote:
> > 
> > Hi folks,
> > 
> > I'm on a quest to find the average circuit-creation rate of clients. I
> > looked in path-spec.txt to find an answer, but it wasn't there. So I
> > thought: lets take some measurements using ARM. This got me some strange
> > results. I start ARM, do some browsing, and close my browser. During the
> > browsing ARM reports 3 circuits with ID's: 398, 399 and 400. These
> > circuits are still there 45 minutes after the browser was closed. If I
> > then restart ARM it reports that there are only two circuits, with
> > circuit ID's 405 and 406. It looks to me that ARM does not update the
> > circuits page when old circuits are closed and replaced by new circuits.
> > It's also possible that ARM keeps old circuits alive after they are not
> > being used anymore by my Tor proxy.
> 
> It seems more likely that this is a refresh issue, either in arm or in
> the tor event code.
> 
> Can you replicate it with the latest stable versions of tor and arm?
> 
> > Note: I use ARM version: 1.4.5.0, Tor version: 0.2.5.12
> 
> That's a very old version of tor. I wouldn't use it to measure anything, we've
> made significant improvements in the last few years.

ARM version 1.4.5 seems to be the latest version. I checked out NYX but
failed to get it running (Unable to load nyx's internal configurations:
[Errno 21] Is a directory: '/home/rob/src/nyx/nyx/settings')

Did test with Tor version 0.2.9.3-alpha-dev. Same problem....

Regards,
Rob.
https://hoevenstein.nl