Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[kuanto]

Hello,

I'm running aMule 2.3.1, and I noticed that the Kad nodes increase periodically without any upper limit: it begins at ~700 on startup, it stays stable for an hour or so, and after that, it increases by +100 nodes. Then, it repeats endlessly this cycle of stability for 1 hour and then increase of +100 nodes. After a few days, it shows ~5500 nodes, and it is still increasing... I'm wondering:

- First of all... what does the "nodes" number exactly stand for?
- How many nodes are normal for Kademlia?
- Is this "endless" increase normal?

I'm wondering how Kad works... I tried to search the forum for some questions, but I only found some very old threads without a clear answer: http://forum.amule.org/index.php?topic=13448.msg88699#msg88699 In this thread, they even suggested that someone could be trying to poison Kad tables! Also, I fear that this endless increase could take too many resources if there is no upper limit.

Could anyone help me understand the Kad behavior  ?

Thanks in advance for your answers !

[HKM]

Currently all my 4 FServ is reporting these KAD nodes.

STHNSv4x1 = 1472
STHNSv4x2 = 1104
STHNSv4x3 = 1472
STHNSv4x4 = 1626

Well i'm not going to lie that number is unusually very high and wouldn't call it normal as I have personally never seen it half of that. I just checked the thread you posted regarding K Bucket poisoning and came across my old post.

I am connected to 879 nodes but it depending on the duration your host is connected to KAD. But normal is 800 to 1100, sometime you might even pass over 1200. Even if you disconnect from KAD it will take 5-6hours before all node removes you from there list and you will get around 6-8 udp queries every 2-3sec.

To add to that as you know nodes.dat only stores 200 nodes which depending on your own KAD hash employs recursive algorithm for node lookup closest to you. As its most important procedure participant must perform which is to locate the closest K nodes to given hash. But you will encounter more nodes depending on ed2k sources, node found through source-exchange and passive basic Kademlia methods as runtime duration increase over time.

EDIT: Removed Link

[HKM]

Well it does look normal, I never looked at it before until now.

STHNSv4x1
Uptime: 1Days 17:33 hours
KAD Nodes: 3312

STHNSv4x2
Uptime: 1 Days 00:44 hours
KAD Nodes: 3864

STHNSv4x3
Uptime: 1 Days 00:39 hours
KAD Nodes: 4416

STHNSv4x4
Uptime: 1 Days 11:45 hours
KAD Nodes: 2122