Congratulations to the Maidsafe team! This is huge 
(sorry for the late reply, but better late than never)
12 Likes
Yes it’s in, you should be given the same XOR-URL while I think it’s still charging you 
7 Likes
Okay, but I didn’t say accept all the spam. I just said that if the goal is to allow general participation then it’s important to keep the barriers to entry low, and that accepting nodes at random would help.
But what about all that spam? Good point! Maybe you’ve already covered this (and if you haven’t, I’m sure you already know it anyway), but a common technique to deal with brute force authentication attacks is to introduce a delayed response to any and all login attempts. That’s not a complete security model on its own, but it’s an example of a technique that has a big impact on network spam without raising hardware and energy costs for everyone else.
Spam has been around for a while. If you wanted, you could come up with a combination of defenses that worked well without making participation impractical for people with modest hardware. But if you’d prefer not to go that route I’m not actually here to fight about it.
In the meantime, I did get this thing running on a raspberry pi model 2 (!) and I’m currently excited about it. Even though I’m too late to join the actual network.
6 Likes
I don’t think you understand the problem. Saying there are techniques to deal with spam (email?) has no bearing on this.
If nodes are selected to join at random, and you have one node trying to join once per second, and I have a thousand nodes trying to join each second. The odds of joining each time are 1000 to 1 in my favour.
3 Likes
In this scenario, you are assuming no workable captcha, right?
Ayr, we have a problem!
We have 3 sections (“00”, “01”, “1”).
The first 2 are OK. Seen from clients their elder lists are:
- “00”:
"071af364","08f1356d","21723a03","22d760d6","2a4277e1","2c83dc26","2d266473" - “01”:
"41ed3b6d","59a171ac","6a9f36b9","6efa2333","72289ca2","75a92c1c","7f0493b4"
But “1” is not OK:
Some clients think that its list is:
"a76176f7","b89ce263","ca246c30","cff01916","d3e291f6","dc0ddac0","e01b9f2a"
But for some others it is:
-
"093e11de","1b5ced41","676222fb","73e7c2fe","77f5e7fa","8a62d444","e78803e9"
(and note that the second list is largely incorrect (only the last 2 belong to this prefix)
Besides this discrepancy section “1” is slower to connect to.
Can @maidsafe investigate this?
EDIT:
I can’t reproduce the section 1 bad list, but I confirm it did happen: a client in this area got once this list. Possibly an intermittent problem.
I also confirm that connexion to section 1 is largely longer than to section 00 or 01 (about 1 minute instead of 6 seconds). It maybe normal because it hasn’t split, so its elders have twice as work to do.
14 Likes
I’m stating the problem, solutions welcome.
Im glad I didnt throw that one away - Time to fish that out of the discarded hardware pile
When I said “network spam” I did not mean “email spam.” C’mon man. If we were working together on solving a problem and you didn’t understand the techniques I was talking about, we’d sit down and talk them over in detail.
This chat is not going in a direction that’s helping either of us. I’m slightly bummed about it.
3 Likes
This is where Anti Entropy kicks in. It’s not in every message yet, but basically it brings things back into sync after churn. So you can sometimes get some nodes (but not >2/3) out of sync, next message (maybe a couple if they need updated ProofChain (the chain part they need to prove the current section key)) gets them back in sync.
12 Likes
Thanks @happybeing and @wydileie for the very useful:
until grep -q "Handling NodeDuty" ~/.safe/node/local-node/sn_node.log; do safe node join; sleep 30; done
I let that run during the night and I now have a node on my macbook!
BTW, I had problems with the instructions to set the path in osx; this worked:
Open up Terminal.
Run the following command:sudo nano /etc/paths
Enter your password, when prompted.
Go to the bottom of the file, and enter the path you wish to add ( /Users/Home/.safe/cli )
Hit control-x to quit.
Enter “Y” to save the modified buffer.
That’s it! To test it, in new terminal window, type:echo $PATH
For what it’s worth, I can confirm the above @StephenC ; I couldn’t join before with my Netgear R7000 but all good this time. Well done again!!
11 Likes
Well you said there were techniques to handle this as if this was obvious without giving any hint of them, something I’ve already said I believe is a hard problem. That sounded like you were brushing this off. Which in turn seems like you either aren’t listening or don’t understand.
If you do great, but your response to my comment about ads and media also seemed naive to me. So that’s the context I’m using to interact with you.
If I’m wrong, and you do have ideas about how to solve this issue why not put them forward. There’s nothing to be gained by letting what I say get in the way of that, and everything to lose.
1 Like
Imo, one way to solve the queue problem is PoUH.
That may not be the best or most elegant way, but it should do the trick. Now all we need to do is figure out a good PoUH. 
4 Likes
It might be, but that’s a hard problem too. If we could achieve that it should increase the cost of scaling the attack, but enough? I don’t know 
1 Like
Should the install.sh script be using the SHELL variable instead of BASH_VERSION / ZSH_VERSION to determine which shell profile to be appending? As I run a zsh shell but the script is modifying my bashrc, presumably as the script is executed by bash it is getting a BASH_VERSION value.
5 Likes
Yes, it would increase the cost substantially. Let’s say for example that each PoUH or PoH takes a human 5 to 15 minutes of actual living work/labor…
People who want to spam would need to hire lot’s of workers to do the manual labor. Ideally the work could be done or performed in the client.
We just need to identify some tasks that only a human can do.
3 Likes
I agree it’s a promising avenue, I don’t think it’s easy though. There are some existing schemes, we’d need to integrate them into SNapp somehow, and the SN itself.
Note I had an insufficient funds error as well, but repeating the command worked, so something weird happening there:
MacBook-Pro-2:~ Home$ safe files put --recursive test
Error: NetDataError: Failed to PUT Public Blob: Transfer(InsufficientBalance)
MacBook-Pro-2:~ Home$ safe files put --recursive test
FilesContainer created at: "safe://hyryyry7rsc19eb1ydtcj8tztr6n68pxisu75r4anc3x3z5k55qgc386jhhnra"
+ test
+ test/.DS_Store safe://hyryyyykszf8567qa98tdsk461m45jf56xgwxmo6x5k4oy8shzcb6c7smwa
+ test/P1010033.jpg safe://hygoygyqisayhextttt5znk8fowh6bx9j4pqxgupcdtmdwiftt3bc896iuc
+ test/P1010046.jpg safe://hygoygyx69nhu8c3b4xcqgi54dibt1rpewz8qg3eq7tx5chb587gj6tst9e
MacBook-Pro-2:~ Home$
@moderators , I thing this thread would be more useful if we stuck to testnet running/joining related issues. Feature requests/ discussions should have a seperate threads IMO, but I might be wrong.
7 Likes
Did we not spend a good couple of years and thousands of posts on PoUH a few years back - and didnt get anywhere? But like Edison said “I also now know 10000 ways NOT to make a light bulb”
1 Like