[Bitcoin-development] Testnet DNS seed

[Bitcoin-development] Testnet DNS seed

[Peter Todd]

[-- Attachment #1: Type: text/plain, Size: 1094 bytes --]

I setup a testnet DNS seed using Pieter Wuille's bitcoin-seeder, with
some simple modifications for testnet. It's at
testnet-seed.bitcoin.petertodd.org I also created
static-testnet-seed.bitcoin.petertodd.org which currently has a single A
record pointing to a testnet node I'm running to bootstrap the seeder
itself.

Everything is running on a dedicated Amazon EC2 micro instance. Just
IPv4 is supported right now as EC2 doesn't support IPv6; even tunnels
are broken. I also haven't setup tor yet. I can do both if there is
demand.

I guess the next step is to create a new strTestNetDNSSeed in the
satoshi client, although it'd be better if at least one more person had
a testnet seed to include in the list. Probably best to leave IRC
enabled too.

Also, FWIW, it looks like the pnSeed list is way out of date...

Pieter: Have you written any start/stop/monitoring scripts for the
seeder? My mods are at git://github.com/petertodd/bitcoin-seeder.git in
the "testnet" branch. I'll send you a pull request once it's had some
testing.

-- 
'peter'[:-1]@petertodd.org

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [Bitcoin-development] Testnet DNS seed

[Jeff Garzik]

On Thu, Jan 24, 2013 at 2:01 AM, Peter Todd <pete@petertodd.org> wrote:
> Everything is running on a dedicated Amazon EC2 micro instance. Just
> IPv4 is supported right now as EC2 doesn't support IPv6; even tunnels
> are broken. I also haven't setup tor yet. I can do both if there is
> demand.

How long do you plan to run this?  Indefinitely [presuming there is
interest and users]?

> I guess the next step is to create a new strTestNetDNSSeed in the
> satoshi client, although it'd be better if at least one more person had

No objection

> a testnet seed to include in the list. Probably best to leave IRC
> enabled too.

Yes, IRC must remain enabled


> Also, FWIW, it looks like the pnSeed list is way out of date...

Yes.  Usually nanotube does the update when poked.  That can happen,
or something different.

A nice alternative might be a simple script that transforms Pieter's
seeds.txt into pnSeed[]



-- 
Jeff Garzik
exMULTI, Inc.
jgarzik@exmulti.com

Re: [Bitcoin-development] Testnet DNS seed

[Peter Todd]

[-- Attachment #1: Type: text/plain, Size: 1198 bytes --]

On Fri, Jan 25, 2013 at 09:23:28PM -0500, Jeff Garzik wrote:
> On Thu, Jan 24, 2013 at 2:01 AM, Peter Todd <pete@petertodd.org> wrote:
> > Everything is running on a dedicated Amazon EC2 micro instance. Just
> > IPv4 is supported right now as EC2 doesn't support IPv6; even tunnels
> > are broken. I also haven't setup tor yet. I can do both if there is
> > demand.
> 
> How long do you plan to run this?  Indefinitely [presuming there is
> interest and users]?

Indefinitely. It's a pretty cheap thing to run, about $7.5/month. If
anyone else wants I can give them a machine image copy easily too.

> > Also, FWIW, it looks like the pnSeed list is way out of date...
> 
> Yes.  Usually nanotube does the update when poked.  That can happen,
> or something different.
> 
> A nice alternative might be a simple script that transforms Pieter's
> seeds.txt into pnSeed[]

seed.txt? You mean the dumpfile produced by bitcoin-seeder? That has
uptime info, although only a months worth if I understand it correctly.

pnSeed probably should be filtered with SORB's dynamic ip list or
similar too, and additionally add an expiry time. (1 year?)

-- 
'peter'[:-1]@petertodd.org

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [Bitcoin-development] Testnet DNS seed

[Jeff Garzik]

On Sun, Jan 27, 2013 at 5:27 AM, Peter Todd <pete@petertodd.org> wrote:
> On Fri, Jan 25, 2013 at 09:23:28PM -0500, Jeff Garzik wrote:
>> On Thu, Jan 24, 2013 at 2:01 AM, Peter Todd <pete@petertodd.org> wrote:
>> > Everything is running on a dedicated Amazon EC2 micro instance. Just
>> > IPv4 is supported right now as EC2 doesn't support IPv6; even tunnels
>> > are broken. I also haven't setup tor yet. I can do both if there is
>> > demand.
>>
>> How long do you plan to run this?  Indefinitely [presuming there is
>> interest and users]?
>
> Indefinitely. It's a pretty cheap thing to run, about $7.5/month. If
> anyone else wants I can give them a machine image copy easily too.

Cool.  ACK on adding your seed to the ref client.

>> > Also, FWIW, it looks like the pnSeed list is way out of date...
>>
>> Yes.  Usually nanotube does the update when poked.  That can happen,
>> or something different.
>>
>> A nice alternative might be a simple script that transforms Pieter's
>> seeds.txt into pnSeed[]
>
> seed.txt? You mean the dumpfile produced by bitcoin-seeder? That has

Correct.

> uptime info, although only a months worth if I understand it correctly.
>
> pnSeed probably should be filtered with SORB's dynamic ip list or
> similar too, and additionally add an expiry time. (1 year?)

That's fine.  The main point was to create something scripted and
repeatable.  Then pnSeed[] becomes easier to maintain _and_ easier to
audit.  Right now it is a bit opaque with a list of hex constants.  A
script or makefile rule could do

     cd src
     ../contrib/gen-seeds.py > seeds.h

and seeds might look like

  #ifndef __SEEDS_AUTOGEN_H__ etc.

  unsigned int pnSeed[] = {
    0xABCD1234, /* IP addr 12.34.56.78 */
  };

I think Satoshi liked them opaque and not easily searchable, so
printing the readable IP address in a comment may not be desired.
Anyway, that output would make updating the list more transparent and
easy to maintain.

-- 
Jeff Garzik
exMULTI, Inc.
jgarzik@exmulti.com

Re: [Bitcoin-development] Testnet DNS seed

[Pieter Wuille]

On Sun, Jan 27, 2013 at 05:27:52AM -0500, Peter Todd wrote:
> seed.txt? You mean the dumpfile produced by bitcoin-seeder? That has
> uptime info, although only a months worth if I understand it correctly.

It has information about all non-banned IPs that were ever connected to
succesfully. The connectability% is a decaying average, with a time 
constant of up to a month, though.

I have a script ready which combines one or more seeds.txt's data into a
pnSeed[] array C source code, by the way. I'll add it to the repository.

-- 
Pieter