[tahoe-dev] Advertised invalid node port

[slush]

Hi,

back to original topic. I checked port on my outside-NAT node again
and there is again bad port given to node by introducer (probably). I
didn't changed anything in configs from time of my first report (4
days ago).

Of course this port is closed (tested using telnet from Internet).

Any idea? Is here any log or something where I can check what happen?
How can I ask introducer to tell me which port it currently know?

Summary (because I sometimes don't understand my own texts ;) :

t + 0 = I found that port of my storage server given to another
(outside-NAT) node is broken. That port is closed, as expected from
firewall rules. I changed config to tell explicit port to bind.
t + 2h = Still bad port
t + ~10h = Port is OK
t + 2-3d = Many restarts of both nodes
t + 4d (now) = Port is wrong again

Thank you for any idea,
Marek

On Wed, Jun 16, 2010 at 1:15 AM, Brian Warner <warner at lothar.com> wrote:
> On 6/15/10 2:48 PM, slush wrote:
>> Update:
>>
>> After few hours and many restarts of both nodes I finally see correct
>> port on slush2. I was confused a little because I expect that new port
>> will be advertised immediately after slush1 restart and reconnect to
>> node slush2.
>
> Hm, it sounds like the Introducer wasn't propagating changes as fast as
> you were expecting. To be honest, I'm not sure how fast it's supposed to
> show up.
>
> The Introducer's web page might be useful (the introducer gets a
> web.port just like client/server nodes). It should list all the
> announcements that it's heard and everything it's currently advertising.
>
> The thing I'm least sure about is what clients do when they get
> "replacement" announcements (i.e. announcements for servers that they've
> already heard about). I remember writing code that spots these and uses
> them to replace the Connector object with a new one (thus initiating a
> new TCP connection), but I don't recall what the already-heard-about
> criteria was. I also remember fixing some of this in my
> #466-signed-introducer work, which makes me think that at least some of
> it was broken beforehand.
>
> You should probably also be aware that tahoe nodes will only initiate
> connections to nodes that advertise themselves as servers. If node A has
> their [storage]enabled config set to False, then other nodes won't try
> to connect to them. (this bit me when I had one client node
> port-forwarded but a second NATed server node was not forwarded, and I
> wanted to take advantage of tahoe's bidirectional connectivity to let
> everybody see the server: I had to mark my client as a server, with
> 'readonly=true' to keep it from accumulating shares, to get all the
> connections to stick around).
>
> cheers,
>  -Brian
> _______________________________________________
> tahoe-dev mailing list
> tahoe-dev at tahoe-lafs.org
> http://allmydata.org/cgi-bin/mailman/listinfo/tahoe-dev
>