Folks, I'm looking for documentation on the high outdegree and inter-UP QRP exchange / negotation / policies. There does not seem to be anything here in the...
A few months ago when developing my Gnutella crawler, I observed that a large percentage of non-leaf nodes refuse connections, by sending a TCP RST packet. I...
Client developers should make sure to stop sending back queryhits when their upload queue becomes full. This sounds scary in theory but works in practice. It...
Windows XP Service Pack 2 has a restriction on simultaneous outgoing TCP/IP connection attempts. So far, it looks like there is no way to modify the...
... Not quite true. There's at least an inofficial patch to remove this limit: http://www.lvllord.de/ After all, you get what you pay for. -- Christian...
In a recent Windows update, I noted that my Wireless WAN config was changed in its firewall setting, by Microsoft which added client support for a service...
... OK, I got no reply whatsoever. I guess you're right. Who needs specifications publicly available anyway? Don't worry: I'll improvise something from my...
Raphael_Manfredi@...
Aug 3, 2004 2:57 pm
20187
What kind of documentation would you like? There's 'Leaf Guided Dynamic Queries', which help leaves to direct ultrapeers when to stop queries. That document...
Raphael, I believe that the document you referenced about dynamic querying contains the information for high outdegree and inter-UP QRP as well. They are ...
... That's the one. Silly me, I was looking under the "QRP" folder, not under "search". Thanks Sam for bothering answering. Raphael...
Raphael_Manfredi@...
Aug 3, 2004 3:21 pm
20190
... Thanks Greg, Yes, as soon as I was pointed to the right place, I found all the documentation I needed. Inter-UP QRP now works fine in GTKG. I see this in...
Raphael_Manfredi@...
Aug 3, 2004 6:38 pm
20191
General question(s) about GUESS out-of-band hits. 1) Here is a typical advertisement of query hits I received, along with my response, both via UDP: Received a...
Version 1 has been depreciated. Use v2. LIME/11v2 I believe that the second byte is the number of desired results. Specs should be updated. Thanks -greg ... ...
Okay, one correction to that email. I see that you were talking about the repliers message not the ack for your second question. The second byte in question...
Hello, Greg, OK gotcha on the 12v2... What is the point of sending a Udp 12v2 with 2nd byte = 0x00 when you cannot receive the 11v1 reply via Udp? Or is a Tcp...
The fact that both are version two is coincidental. We upped LIME/11 to v2 almost at the very start of using OOB (quite a while ago), because it was necessary...
[Quote] LimeWire will not accept a LIME/11v1 'cause we were the only people using it at the time, and it made no sense to do backwards compatibility for...
Hi everybody, I need to simulate a Gnutella network in a network simulator to prove some theories for my thesis. Now instead of writing just another Gnutella...
... Sorry, it's me again. The documentation for the inter-UP QRP exchange does not mention whether nodes advertising: X-Ultrapeer-Query-Routing: 0.1 are...
Raphael_Manfredi@...
Aug 3, 2004 10:52 pm
20200
1 bit QRP patches have nothing to do with Ultrapeer Query Routing. It is an optimization that is more or less unnecessary, because the entire message is going...
... Ah, thanks for the quick information. BTW, my tests show that for a QRP table that is about 50% full, there is a difference in FINAL SIZE between 4-bit and...
Raphael_Manfredi@...
Aug 3, 2004 11:23 pm
20202
That bandwidth level for 300 leaves doesn't make sense. I would think that the aggregate bandwidth for the ultrapeer would be very high. Thanks -greg ... ...
... No, I remade the measurements this morning. On AVERAGE, (exponential moving) the bandwidth was 4.5 K/s in and 5.5 K/s out. What matters is more the...
Raphael_Manfredi@...
Aug 4, 2004 8:59 am
20204
From: "Sam Berlin" <sberlin@...> ... Did you forget my source code patch (about 6 months old) that allowed interpreting incoming 1-bit patches, and...
From: "Greg Bildson" <gbildson@...> ... You're right! What will eat the most the bandwidth of Ultrapeers is not the number of leaf nodes, or their own...
... That reduces Upload pressure a little bit, by not inviting more new clients to pound away at a full Upload queue. But it does not address at all the fact...
Here's the deal. If most frequent keywords are encoded in the QRP table then the fill ratio is irrelevant. If you have matches for 50% of popular keywords...