Anonymous Servers - stats/latency question

Author

stats/latency question

Nomen Nescio

2007-08-21, 1:13 am

I've been wonndering about this for quite some time and have
reached the point where I have to risk the flames and ask.

Why do latency times vary from remailer to remailer?

For example...

Remailer 'a' says it takes 30 minutes to get a ping back from
remailer 'b'.
Remailer 'c', however, says it takes 4 hours to get a ping back
from remailer 'b'.

Since I seriously doubt that remailer 'b' is sitting there
going, "This is from remailer 'c', so I'm gonna ignore it for a
while", I'm at a loss to explain why this happens.

I understand that reliable holds messages in a queue for a
random amount of time, but what I'm seeing is consistant - not
random. Remailer 'c' _ALWAYS_ takes 4 or 5 hours to get a reply
from remailer 'b' and remailer 'a' _ALWAYS_ gets a reply from
remailer 'b' in a matter or 30 to 40 minutes.

I'm guessing that this phenomenon has something to do with why I
have been unable to successfully create a nym on ANY nym server
in two weeks of trying and why my APAS posts only show up about
1/2 the time, and then in duplicate.

Anonymous Remailer (austria)

2007-08-21, 7:14 am

> I'm guessing that this phenomenon has something to do with why I
> have been unable to successfully create a nym on ANY nym server
> in two weeks of trying and why my APAS posts only show up about
> 1/2 the time, and then in duplicate.
I don't thimk so, though that's a legitimate question. If you're a
Windows user inexperienced in handling anon remailers you should start
with an integrated package and step by step migrate to the system you
aim at. I run OmniMix and get nearly all messages through no matter
whether a nym server is involved or not, though sometimes it's a
mystery to me why even with many copies it takes hours and hours and
hours. The reason for losing newsgroup postings it's mostly the
mail2news gateway (too many newsgroups, message too large).

Cheers
..

Zax

2007-08-21, 7:14 am

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On Tue, 21 Aug 2007 07:50:36 +0200 (CEST), Nomen Nescio wrote in
Message-Id: <363d4d7a25a2c08f1cfe5920fb2f3708@dizum.com>:

> I've been wonndering about this for quite some time and have
> reached the point where I have to risk the flames and ask.
>
> Why do latency times vary from remailer to remailer?
>
> For example...
>
> Remailer 'a' says it takes 30 minutes to get a ping back from
> remailer 'b'.
> Remailer 'c', however, says it takes 4 hours to get a ping back
> from remailer 'b'.

I think you're confusing remailers and pingers. Pingers compile
statistics on remailers. Can you cite an example where you're seeing
such huge differences as 30mins vs. 4hrs? It's normal for there to be
some variation, that's just good entropy on behalf of the remailer, but
it shouldn't be that broad.

> Since I seriously doubt that remailer 'b' is sitting there
> going, "This is from remailer 'c', so I'm gonna ignore it for a
> while", I'm at a loss to explain why this happens.

Possible but very unlikely. It would be easy for an operator to skew
latency results for specific pingers if they felt so inclined. Hard to
imagine a motive for it though.

> I'm guessing that this phenomenon has something to do with why I
> have been unable to successfully create a nym on ANY nym server
> in two weeks of trying and why my APAS posts only show up about
> 1/2 the time, and then in duplicate.

At the moment all pingers are providing good information. At times in
the past this hasn't been the case with some old timers advertising
stats that were in some cases years out of date. Providing you're
actually using a current pinger there really shouldn't be an issue.

With regard to creating a nym, how are you going about it? I'm a strong
believer in hand-rolling a simple nym to begin with. This gives you a
feel for how the process works. Sure, tools like QS and JBN simplify
the creation, but in doing it they also cast an opacity over what's
really happening.

How about creating a nym with this reply-block:-

Reply-Block:
::
Remail-To: me@my.address.com

It's valid, relies on no remailers and enables you to communicate
bi-directionally with the nymserver. It offers no anonymity, but
provides an excellent starting point to build from.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGyraUlKZ6CY7Vd0MRClLhAKC5NIeEpZUf
zpEgXa5WFgEFrNRWiwCeKoSA
d7sHFe7QJWjD0QD03EOvHwQ=
=h1BV
-----END PGP SIGNATURE-----

--
pub 1024D/8ED57743 2003-07-08 Bananasplit Operator
Key fingerprint = 796F 67E0 E890 A0BB BDAE EBB4 94A6 7A09 8ED5 7743
uid Admin <admin.bananasplit.info>

George Orwell

2007-08-21, 1:13 pm

Zax wrote:

>With regard to creating a nym, how are you going about it? I'm a strong
>believer in hand-rolling a simple nym to begin with. This gives you a
>feel for how the process works. Sure, tools like QS and JBN simplify
>the creation, but in doing it they also cast an opacity over what's
>really happening.

It's always a positive thing to know what's going on in such a
sensitive sector. But how many email traffic would you expect to be
there if every prospective user in the past had been recommended to
understand at least RFCs 2821, 2822, 1939, 2045, 2046 first?

Zax

2007-08-21, 1:13 pm

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On Tue, 21 Aug 2007 14:25:59 +0200 (CEST), George Orwell wrote in
Message-Id: < 64a2f2cbc19fd13bad2a725e490c6466@mixmast
er.it>:

> It's always a positive thing to know what's going on in such a
> sensitive sector. But how many email traffic would you expect to be
> there if every prospective user in the past had been recommended to
> understand at least RFCs 2821, 2822, 1939, 2045, 2046 first?

For sure there would be less email users but the remaining ones would
probably be more capable of diagnosing any issues they encounter.
Reading RFC's is going to extreme, but a basic understanding of how mail
works would take most users a couple of hours and they would benefit
greatly from the effort.

Even without the user effort, there are thousands of ISP helpdesks out
there that can help with an Outlook Express problem but there are few
contacts for "my Nym doesn't work". :-)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGyvFVlKZ6CY7Vd0MRCrVcAJ0Zbyb6knnB
zIMj5OdzvqeX0QivtQCg9t/J
63PRVjVG/HRGzBdjEPwH6E8=
=pM4u
-----END PGP SIGNATURE-----

--
pub 1024D/8ED57743 2003-07-08 Bananasplit Operator
Key fingerprint = 796F 67E0 E890 A0BB BDAE EBB4 94A6 7A09 8ED5 7743
uid Admin <admin.bananasplit.info>

Cyberiade.it Anonymous Remailer

2007-08-21, 1:13 pm

Zax wrote:

>On Tue, 21 Aug 2007 14:25:59 +0200 (CEST), George Orwell wrote in
>Message-Id: < 64a2f2cbc19fd13bad2a725e490c6466@mixmast
er.it>:
>
>
>For sure there would be less email users but the remaining ones would
>probably be more capable of diagnosing any issues they encounter.

Shouldn't we aim at making anonymous messaging a mass movement? :-))

>Reading RFC's is going to extreme,

So you wouldn't see the construction of nym server control messages,
which includes arranging mail headers and additionally being nimble in
encryption methods, as a task much more demanding than that? I could
even imagine computer experts failing to build nym messages.

> but a basic understanding of how mail
>works would take most users a couple of hours and they would benefit
>greatly from the effort.

Of course it would help. But being realistic, such a recommendation to
a newbie usually is counterproductive in that it mostly deters the
interested person opposite. IMHO a step-by-step approach beginning
with an automated setup would be more successful by avoiding
frustration. Maybe later on a positive feedback arouses interest to
research into the basics. But if that's mandatory for "simply" sending
and receiving nym messages, something's wrong with it on principle.

>Even without the user effort, there are thousands of ISP helpdesks out
>there that can help with an Outlook Express problem but there are few
>contacts for "my Nym doesn't work". :-)

That's why Zax over there at a.p.a-s is sooo important. ;-)
..

Cyberiade.it Anonymous Remailer

2007-08-21, 1:13 pm

Nomen Nescio <nobody@dizum.com> wrote:

> I'm guessing that this phenomenon has something to do with why I
> have been unable to successfully create a nym on ANY nym server
> in two weeks of trying and why my APAS posts only show up about
> 1/2 the time, and then in duplicate.

It's not just you. Nyms are a bastard to set up properly.

NOT impossible, but not easy!

Nomen Nescio

2007-08-21, 7:14 pm

In article <faecqk$v0d$1@bananasplit.info>
Zax <admin@bananasplit.info> wrote:
>
>
> I think you're confusing remailers and pingers. Pingers compile
> statistics on remailers.

Whichever. Remailers also ping other remailers to gather stats.
Whether the stat source is a remailer or a pinger is a
difference which, in this case, makes no difference.

> Can you cite an example where you're seeing
> such huge differences as 30mins vs. 4hrs?

That was an example that I pulled out of my XXX. However, here
is what I just found by comparing stat sources listed at
http://www.noreply.org/allpingers/

http://www.panta-rhei.eu.org/stats/stats/rlist2.html has the
following entry for frell remailer:

frell ?????8793A77 7:09 ?????++++++7 96.2% PR GO
ATLEUIN9

However, http://stats.mixmaster.it/rlist2.html shows

frell 426653554364 3:42 ++++7++++++9 98.7% PR GO
ATLEUIN9

and http://rlist.ath.cx/rlist2.html shows that remailer with
much higher reliability than Panta does

frell 645567455554 4:12 +++++++++++8 99.3% PR GO
ATLEUIN9

So my question is, why does Panta show an average of just over 7
hours to get a reply to a ping while George and Kroken get a
reply in 3.5 to 4 hours? And, why does Panta show a lower
reliability percentage? I understand that the question marks in
Panta's stats mean that he's working with 7 days' data instead
of 12, and that *MAY* account for the reliabily difference, but
it shouldn't, in my mind, almost double the ping times.

> At the moment all pingers are providing good information.

Really? These are listed at http://www.noreply.org/allpingers/

http://bikikii-remailer.noneto.com/stats/rlist2.html Generated:
Sun 13 May 2007 07:16:53 GMT
http://www.cside.dyndns.org/stats/rlist2.html Generated: Fri 17
Aug 2007 18:00:04 GMT
http://dingo.1hwy.com/Dingorlist2.html Generated: Tue 6 Mar 2007
21:54:22 GMT
http://www.gurski.org/remstats/rlist2.html The requested URL
/remstats/rlist2.html was not found on this server.

I'm guessing that www.noreply.org isn't quite as accurate as the
owner would like everyone to believe. On the page I cited above,
it says (below the list of pingers):

>The canonical list of all pingers in a machine readable format
can be found here:
>allpingers.txt. It was last updated Thu Feb 22 14:04:20 2007.

If the 'canonical list of all pingers' is that far out of date,
how does one get accurate stats to work with? Trial and error?

>
> With regard to creating a nym, how are you going about it? I'm a strong
> believer in hand-rolling a simple nym to begin with. This gives you a
> feel for how the process works. Sure, tools like QS and JBN simplify
> the creation, but in doing it they also cast an opacity over what's
> really happening.

I've hand-rolled nyms before. Successfully, even. It can be a
major PITA, but I do understand how the process works.

I've sent several config messages with QuickSilver and one or
two with JBN. I've actually gotten three confirmation messages,
but my replies (to activate the reply blocks) seem to be going
into a black hole somewhere between me and the nym server.

Nomen Nescio

2007-08-21, 7:14 pm

Zax

2007-08-22, 7:14 am

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 21 Aug 2007 18:33:37 -0000, Nomen Nescio wrote in
Message-Id: <5WN94ZMU39315.8566782407@anonymous.poster>:

> So my question is, why does Panta show an average of just over 7
> hours to get a reply to a ping while George and Kroken get a
> reply in 3.5 to 4 hours? And, why does Panta show a lower
> reliability percentage? I understand that the question marks in
> Panta's stats mean that he's working with 7 days' data instead
> of 12, and that *MAY* account for the reliabily difference, but
> it shouldn't, in my mind, almost double the ping times.

You've selected probably the best example of a high variance in latency.
Frell is a high-latency node, averaging around 4 hours per message.
This high latency is achieved by configuring a very large pool size.
The pool is periodically read and a defined percentage of messages
within it are processed. Because the pool is so large, many pings
remain outstanding for a long period of time which causes the variation
in reported latency. If you look at
http://stats.mixmin.net/frell.godot...rell.eu.org.txt you'll see
that virtually every pinger has a number of pings outstanding from it.

>
> Really? These are listed at http://www.noreply.org/allpingers/

Sorry, I should have been more specific. All the pingers that are
listed on the metastats http://stats.mixmin.net should be returning good
results. I'll contact the maintainer of allpingers and advise him that
dingo and bikikii are now dead. Hopefully cside and gurski will be back
in operation soon.

> If the 'canonical list of all pingers' is that far out of date,
> how does one get accurate stats to work with? Trial and error?

I'd suggest checking out the metastats page. Perhaps I'll try and pull
together a page that compares each pinger against the others so that
users can see at a glance which are providing good results.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGzAe9lKZ6CY7Vd0MRCl7OAKDuRMdFbouR
wRaMFZsZB4hfVJdpgACdEnTo
ff7bn49M/tYJ9w+9UsRwndo=
=Cyli
-----END PGP SIGNATURE-----

--
pub 1024D/8ED57743 2003-07-08 Bananasplit Operator
Key fingerprint = 796F 67E0 E890 A0BB BDAE EBB4 94A6 7A09 8ED5 7743
uid Admin <admin.bananasplit.info>