Re: [Help-gnunet] bandwidth on ADSL

[dot]

Christian wrote:

>Should definitely not happen. Do you recall how many connections your node

>established?  What did you set your upstream limit to be?

Pretty much the same 16 nodes were connected all the time.

I wrote a script to grab the IPs and hostnames, and they were pretty much
always connected.

>The HELOs do not contain the BPS limits, so no.  What I think might be
going 
>on is that the traffic comes from plenty of peers who try to connect to
you 
>and who push you above the limit before you can tell them to 'slow down'
to 
>200 bps.  In other words, peer X learns about you, and just 'tries' to
send 
>you a message.  Since that is successful (you have high-upstream 
>availability, so you always reply, and your down-stream limit is
artificial 
>so you always receive) X tries to connect.  If many peers do this, this
alone 
>might push you above 200 bps. 

Does that mean I have to make the node "look" less responsive to other
nodes?

Would that mean they'd assign me less credit for not replying as much, or
more credit, for not sending them as much crap?

>From what I've been able to find (in the various shaping HOWTOs) there's no
way to really control inbound packets in anything but TCP, and even that's
only by setting upstream shaping and hoping the other end reciprocates.



>gnunet-stats output should tell us more.

here's the output of gnunet-stats after running for about 5 minutes
(hopefully email won't wrap lines):

$ gnunet-stats|sort
# blocks AFS storage left (estimate)                        :         
2986283
# bytes decrypted                                           :          
346684
# bytes noise sent                                          :          
266184
# bytes of noise received                                   :           
52016
# bytes received and decryption failed                      :              
 0
# bytes received of type 0                                  :           
12408
# bytes received of type 1                                  :           
11440
# bytes received of type 16                                 :          
257828
# bytes received of type 18                                 :           
33924
# bytes received of type 2                                  :             
420
# bytes received of type 3                                  :             
308
# bytes received of type 5                                  :            
1912
# bytes received of type 6                                  :           
52016
# bytes received of type 8                                  :             
276
# bytes received via tcp                                    :          
305600
# bytes received via udp                                    :           
68528
# bytes sent via tcp                                        :          
565472
# bytes sent via udp                                        :           
93384
# bytes transmitted of type 0                               :           
45684
# bytes transmitted of type 1                               :           
40040
# bytes transmitted of type 16                              :          
441560
# bytes transmitted of type 18                              :          
460544
# bytes transmitted of type 2                               :            
1232
# bytes transmitted of type 3                               :             
392
# bytes transmitted of type 5                               :            
6520
# bytes transmitted of type 6                               :          
266184
# bytes transmitted of type 7                               :             
120
# bytes transmitted of type 8                               :             
636
# currently connected nodes                                 :              
12
# encrypted bytes sent                                      :         
1179444
# kb dupe content in                                        :              
 2
# kb ok content in                                          :              
15
# kb orphan or pushed content in                            :              
16
# lookup (3HASH, search results)                            :              
 0
# lookup (CHK, inserted or migrated content)                :             
436
# lookup (ONDEMAND, indexed content)                        :              
 0
# lookup (SBlock, search results)                           :              
 0
# lookup (data not found)                                   :              
 0
# messages expired (bandwidth stressed too long)            :             
117
# messages in all queues                                    :             
502
# p2p CHK content received (kb)                             :              
32
# p2p SBlocks received                                      :              
 0
# p2p namespace queries received                            :              
 0
# p2p queries received                                      :            
2320
# p2p queries sent                                          :            
4097
# p2p search results received (kb)                          :              
 0
# p2p super queries received                                :            
2317
# routing table entry already in place                      :             
584
# routing table entry replaced                              :            
2236
# routing table full                                        :              
53
# times outgoing msg deferred (bandwidth stressed)          :            
1023
# times outgoing msg sent (bandwidth ok)                    :             
815
% of allowed cpu load                                       :              
25
% of allowed network load (down)                            :             
490
% of allowed network load (up)                              :              
57
Uptime (seconds)                                            :             
322
$

Five minutes' worth is probably not entirely representative. For example, %
of allowed down is not usually 490%. 

In fact when gnunetd starts, DOWN is 0, and UP is over 100000, then they
both go to about 30000, then DOWN hovers at a few thousand while UP is less
than 20.

...I'll upload another gnunet-stats later when it's run a bit longer and
stabilised itself.

Something else just occurred to me: I have migration enabled. 
Could all this traffic just be nodes offering things for me to store?

Sorry if this sounds uninformed. I know there's protocol documents for
ESED2 and TCP (and should probably RTFM :-)

Thanks for your help.