14 Jun
2017
14 Jun
'17
6:49 a.m.
> I think we will have to agree to disagree about this.
Ok, let's focus on the problem first.
Conclusions can be made later.
> Please help us find out which of these things impact one of your
> relays.
The only thing from your list, which can have effect - is the properties
of other relays and theirs Internet connections.
But weight is assigned to single relay, so this differences must be
filtered out somehow.
> Yes, the Tor network measures it from 4 different locations every few
> days.
And gives incorrect weight as a result.
Stable incorrect result.
It is needed to discuss what is a good weight.
But that can be done later.
My idea is that 1 KiB of load when 1000 KiB of bandwidth available is bad.
Just imagine that this 1 MiB/s are really can be used.
Of course, not to all network.
But if some dial-up node can't use it, this doesn't mean that no one can.
> What makes your measurement is more accurate?
1. I was checking this speed with random relays, which mean that
high-speed relays was certainly included in the circuits.
(this gives 1000000 instead of 58436)
2. I was used obtained value directly to decide if relay is good enough.
(this gives 1000000 instead of 5000)
> Where are you measuring from?
Ukraine.
> Is it close to the relay?
I have made many measurements.
Some of them was close to relay, some are not.
> How long did it take to do the download?
Here are some results:
extendcircuit 0 $BD4354E76929C90B7004FF149A3C52189A3B4634,$A53C46F5B157DD83366D45A8E99A244934A14C46
650 CIRC 10 BUILT $BD4354E76929C90B7004FF149A3C52189A3B4634~Hedgehog,$A53C46F5B157DD83366D45A8E99A244934A14C46~csailmitexit PURPOSE=GENERAL TIME_CREATED=2017-06-14T06:02:21.529447
650 STREAM 13 CLOSED 10 38.229.72.16:443 REASON=DONE
$ curl --socks5-hostname localhost:9050 --insecure -O https://38.229.72.16/bwauth.torproject.org/16M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 16.0M 100 16.0M 0 0 856k 0 0:00:19 0:00:19 --:--:-- 985k
But this relay is close to my location.
Let's select another:
extendcircuit 0 $38BF40B902ABC23B4E1503BE9131F1A3BF8EBAC5,$A53C46F5B157DD83366D45A8E99A244934A14C46
650 CIRC 11 BUILT $38BF40B902ABC23B4E1503BE9131F1A3BF8EBAC5~bzerorelay1,$A53C46F5B157DD83366D45A8E99A244934A14C46~csailmitexit PURPOSE=GENERAL TIME_CREATED=2017-06-14T06:06:19.551317
650 STREAM 17 CLOSED 11 38.229.72.16:443 REASON=DONE
$ curl --socks5-hostname localhost:9050 --insecure -O https://38.229.72.16/bwauth.torproject.org/16M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 16.0M 100 16.0M 0 0 268k 0 0:01:01 0:01:01 --:--:-- 271k
Of course, it is slower.
But still far from utilized 2 KiB/s.
> Did you measure from different parts of the world?
I don't have control over multiple locations.
> 1. Choose a relay you control to focus on.
> 2. Send information about the relay's CPU and RAM and configured
> connection limit.
* CPU: Intel Core i5-4690
* RAM: Team Group Dark-1600, 8 GiB
* RelayBandwidthRate 1 MBytes
RelayBandwidthBurst 3 MBytes
> 3. Measure the actual connection limit, bandwidth and latency from
> the rest of the Tor network. (Or from at least 2 locations in the
> US and Western Europe.)
I doubt if it is good to test with a loop connections, so I will post
speedtest.net results. They give good estimate of my connection
properties.
http://www.speedtest.net/my-result/6375679554
DOWNLOAD 94.86Mb/s
UPLOAD 95.21Mb/s
PING 24 ms
SERVER SAINT PETERSBURG
http://www.speedtest.net/my-result/6375676443
DOWNLOAD 89.53Mb/s
UPLOAD 94.51Mb/s
PING 48 ms
SERVER MILAN
http://www.speedtest.net/my-result/6375670157
DOWNLOAD 89.53Mb/s
UPLOAD 94.54Mb/s
PING 51 ms
SERVER DRESDEN
http://www.speedtest.net/my-result/6375655269
DOWNLOAD 85.01Mb/s
UPLOAD 24.01Mb/s
PING 131 ms
SERVER NEW YORK CITY, NY
http://www.speedtest.net/my-result/6375666714
DOWNLOAD 53.70Mb/s
UPLOAD 16.19Mb/s
PING 205 ms
SERVER SAN FRANCISCO, CA
http://www.speedtest.net/my-result/6375685977
DOWNLOAD 24.69Mb/s
UPLOAD 13.46Mb/s
PING 291 ms
SERVER TOKYO
Maybe Atlas graphs also can help:
https://s8.hostingkartinok.com/uploads/images/2017/06/caaccae1a967a838871cdea5739e9b7d.png
https://s8.hostingkartinok.com/uploads/images/2017/06/60ff981f51002bd177e8c991748205cf.png
> Or:
> Change they relay's keys, wait a few weeks, and let us know if
> the bandwidth measurement is better or worse.
I don't want to lose the state, which reproduces the bug.
> If it is better, then the relay was put in a low bucket, and was stuck
> in that bucket. This can happen at random, or if the relay was slow in
> the past.
My relay was never slow.
Possibility of such random stuck is a thing, which is needs to be eliminated.
-- Vort