[Mldonkey-bugs] [bugs #9764] bandwidth waste: stupid response on Sources Answer

[anonymous]

This mail is an automated notification from the bugs tracker
 of the project: mldonkey, a multi-networks file-sharing client.

/**************************************************************************/
[bugs #9764] Latest Modifications:

Changes by: Anonymous user
Date:  
                sam 21.08.2004 at 17:17

------------------ Additional Follow-up Comments ----------------------------
Since those IPs come from what should be a Direct_address list, it shows how 
much crap is being kept...
IMHO a best fix would be to prevent them from being added to sources lists in 
the first place...







/**************************************************************************/
[bugs #9764] Full Item Snapshot:

URL: <http://savannah.nongnu.org/bugs/?func=detailitem&item_id=9764>
Project: mldonkey, a multi-networks file-sharing client
Submitted by: 0
On: dim 25.07.2004 at 20:42

Category:  eDonkey-plugin
Severity:  5 - Average
Item Group:  Program malfunction
Resolution:  None
Assigned to:  None
Status:  Open
Release:  2-5-21
Release:  
Platform Version:  Linux
Binaries Origin:  CVS / Self compiled
CPU type:  Intel x86


Summary:  bandwidth waste: stupid response on Sources Answer

Original Submission:  MlDonkey wastes bandwidth with stupid responses to emules 
extension. "File Request Answer (0x85) [0xc5 <int msgsize> 0x85]" often gives 
out a 
"ip: 0.0.0.0; port: 0" konfiguration, multiple times.

from ethereal:
00e0  80 04 ->> c5 8b 05 00 00 82<<--  6e d2 d2 57 c0 d4 77 f4   ........ 
n..W..w.
00f0  93 70 d4 e9 05 a4 2d ad  32 00 d5 a8 68 e8 36 12   .p....-. 2...h.6.
0100  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00   ........ ........
0110  00 00 00 00 00 00 50 90  a2 5a 36 12 00 00 00 00   ......P. .Z6.....
0120  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00   ........ ........
0130  00 00 52 30 17 f1 39 12  00 00 00 00 00 00 00 00   ..R0..9. ........
0140  00 00 00 00 00 00 00 00  00 00 00 00 00 00 d9 5e   ........ .......^
0150  b9 f2 36 12 00 00 00 00  00 00 00 00 00 00 00 00   ..6..... ........
0160  00 00 00 00 00 00 00 00  00 00 53 68 2e da 36 12   ........ ..Sh..6.
0170  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00   ........ ........
0180  00 00 00 00 00 00 52 50  c1 36 1e 16 00 00 00 00   ......RP .6......
0190  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00   ........ ........
01a0  00 00 53 1a f2 dc 36 12  00 00 00 00 00 00 00 00   ..S...6. ........

Follow-up Comments
------------------


-------------------------------------------------------
Date: sam 21.08.2004 at 17:17       By: 0 <None>
Since those IPs come from what should be a Direct_address list, it shows how 
much crap is being kept...
IMHO a best fix would be to prevent them from being added to sources lists in 
the first place...


-------------------------------------------------------
Date: mer 11.08.2004 at 06:52       By: 0 <None>
well, regarding parsing the packets, I rely on ethereal (and my little 
knowledge about edonkey protocol). 
but I can see the values in the hexpacket, so I am sure, they are shown right. 
The packets have been captured with tcpdump (snapsize 1024 byte), so think I 
also can rely on that..

I checked back.
I am downloading a file, with 19 chunks (said by files.ini), 2 found sources 
(said by file_sources), and got this attached packet on wire.

there are 40 IP's in that packet.

some 0.0.0.0 IP's have a port assigned
some IP's in there are without a port
some IP's are just illegal (Network-addr, start with 0.0, etc. p.p.)

I have force_high_id enabled .. I have a direct connection to the net .. 
shouldn't the whole packet then be 0.0.0.0 ?




-------------------------------------------------------
Date: lun 09.08.2004 at 13:47       By: mldonkey admin <mldonkey>
Are you sure of your method to parse the packets ? Message 0x82 in 
sourceexchange > 1 is supposed to contain, for each source, its IP address and 
port, the IP address and port of the server (or zeros if HIGH-ID), and the UID 
of the source (which is not known by mldonkey since it doesn't keep this 
information, thus sending a lot of zeroes again).


-------------------------------------------------------
Date: lun 26.07.2004 at 09:08       By: 0 <None>
its even worse than i thought.
bash-2.05b$ grep IP /tmp/networkbug | sort | uniq -c
     28         IP: 0.0.0.0 (0.0.0.0)
      1         IP: 0.0.217.157 (0.0.217.157)
      1         IP: 0.0.217.235 (0.0.217.235)
      1         IP: 0.0.82.80 (0.0.82.80)
      1         IP: 146.16.0.0 (146.16.0.0)
      1         IP: 210.192.25.82 (210.192.25.82)
      1         IP: 217.234.39.184 (217.234.39.184)
      1         IP: 24.196.54.18 (24.196.54.18)
      1         IP: 4.92.54.18 (4.92.54.18)
      2         IP: 54.18.0.0 (54.18.0.0)
      1         IP: 62.147.213.190 (62.147.213.190)
      1         IP: 68.157.54.18 (68.157.54.18)

28 IP's with 0.0.0.0 (nonsense)
3 IP's with 0.0.x.x (nonsense)
2 IP's with x.x.0.0 (class B network address = nonsense)

makes 33 from 40 IP's which are nonsense.

Suggested QuickFix: just send every addr/port-pair once.









File Attachments
-------------------

-------------------------------------------------------
Date: mer 11.08.2004 at 06:52  Name: details  Size: 8,43KB   By: None
ethereal packet log
http://savannah.nongnu.org/bugs/download.php?item_id=9764&amp;item_file_id=1575






For detailed info, follow this link:
<http://savannah.nongnu.org/bugs/?func=detailitem&item_id=9764>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.nongnu.org/