NAS :: Descendant of NXT 100% PoS|More Exchanges|WinNAS is ready!

[patapato]

Blacklisting lcharles node might be a bit extreme.  If it was done by lcharles setting his node then it probably wasn't intentional. 

He recognizes explicitly that he restarts his node before the end-of-leasing block 278461, and he restarts it to a time after the beginning of leasing, with a copy made after he started to forge with leased balance. That is an intentional act, or better said, two intentional acts.

The non intentional act is to let the node evolve by itself without reseting. Or at least, to reset his node to scratch, deleting his blockchain file in order to reconstruct a clean blockchain from the network. But reseting to a state between the 1440 blocks of his leased forging power is an intentional act.

[1714610725]

1714610725

[1714610725]

1714610725

[1714610725]

1714610725

[tiker]

Well, he's on the forums now so I'll let him defend himself...

If we blacklist his IP, what do you expect will happen?  My nodes seem to disagree with the blocks your node has.  Chances are, your node will disagree with what my nodes have.  User1234567890 reset a node and ended up where my nodes sit today.  Does that make my nodes the correct chain even though they're stuck at a block lower than your nodes?

[patapato]

Your node disagrees with mine when your block height is over 277934. But it goes back to that height from time to time, and that block height is in agreement with my node. Your node disagrees just when there is a next 277935 block forged by lcharles123 node, or forged by you and confirmed by lcharles123 node.

My node is also stuck because there are no others confirming blocks higher than 279189, who was forged by User1397367406. So it will wait until some other forger reaches that height (that forger was User1397367406, but he reseted his blockchain confused by your case, sorry to say).

If User1397367406 also blacklist lcharles123 node, maybe your and his nodes agree on a different fork than me, but the network will synchronize by itself when you both reach my block height.

What is not healthy, and not logical (it is a weakness in NXT model of leasing) is that 2 nodes are forging at the same time with the same balance, as it is the present case. I mean, leasing period should be a time period, not a block height period, in order to avoid such a situation. Now we have to avoid that situation by ourselves, "by hand", and it should be by understanding all of you that, when somebody lease his balance for short time, it is for short TIME. If all of you agree with that, you should agree that the intended lease period should have already ended. The way to stop the unintended leasing period is to blacklist the unintended forger.

[tiker]

Since Patapato has Pidgin installed to chat, we're going to discuss things in a chat room, (faster than message boards).

This is an open invitation to everyone who wants to join, the XMPP (Jabber) chat room (MUC) is:  nas-dev@conference.funkymonkey.org

Not sure when we'll be discussing yet.. but join and sit idle for now... hopefully everyone with a NAS stake will join.  We need to make a decision to fix this issue soon.

[patapato]

Well, the problem, as I understand, is that lcharles123 has the power of forging with my balance for 1440 blocks, and his fork entered in a loop without end.

Maybe it is not correct to ask others for blacklisting anybody. In that case, the solution by myself will be based on the same thing than the problem, that is, the fact that my forging power is > 80%. So, here is what I can do:

• I setup another node, while keeping my original node intact
• I blacklist my original node patapato.ddns.net, and put nodes of lcharles123 and tiker as wellKnownPeers, in order to ensure that my new node sync with the other fork
• I wait to fully synchronize the forked blockchain
• I wait for the blockchain going back to the block 277935, the block in which lcharles123 forging power begins, as this fork is repeatedly going back to that block height
• In that client, in that moment, I transfer all my 22 M Nas stake to another account of myself, in order to take out all the forging power from lcharles123 in that fork
• I stop forging on my original client, in order to not conflict and let the block height grow in the forked chain
• I wait to the end of the leasing on that fork, at block height higher than 278461
• I restart forging on my original client (on this blockchain I didn't transfer my stake)

In that moment, the loop in that fork will be over, and I will not forge on it. So, it will be a fork with much less forging power and much lower block height. Anyway, the network should resync to one of both forks, and this time, if any forger try to go back with an old saved blockchain file he will not success.

[User1397367406]

To worsen things, a trusted forger thought that he was on a bad fork, so he tried to reset his blockchain, stopping forging on his original blockchain and getting in sync with the attacker's fork:

I tried syncing a client on my computer from mid march and so far it seems to sync until block 277934 and 277940 :
...

I tried to sync a client on my main computer, different from my forging node, and ended up with the above mentioned result. I wanted to see what would happen.
I re-started my forging node yesterday and stopped forging on it, it's still there doing nothing, at block 280284   4/7/2015 22:19:56, connected to 4 nodes: funkyvps3, patapato, 99.227.*.* and 106.187.*.*
The blockchain was not reset but i was forging all the block and so maybe i'm on my own fork.

Solutions?:

• I ask you, @User1397367406, to please recover your last original blockchain, as you forged in sync with me at least until height 279189 (06/04/2015 21:10:52, GMT+2), while you are currenly going back to 277934.
  
• I ask every forger to blacklist lcharles node: nxt.servebeer.com (46.105.44.4)
  
• I will distribute my account balance in smaller accounts, roughly the size of the next big forger, in order to have a more equilibrated distribution of forging power
  

Other suggestions, comments, ideas?

I'll blacklist lcharles node. Not sure if i ever saw his node connected. Should start taking screenshots of the connected nodes and such
I'll try connecting my other client to see what happen, enabling some logging.

I agree on only distributing/leasing a smaller amount of NAS.

[patapato]

Edit: not correct post, sent very late with obfuscated mind, sorry.

[patapato]

I tried to sync a client on my main computer, different from my forging node, and ended up with the above mentioned result. I wanted to see what would happen.
I re-started my forging node yesterday and stopped forging on it, it's still there doing nothing, at block 280284   4/7/2015 22:19:56, connected to 4 nodes: funkyvps3, patapato, 99.227.*.* and 106.187.*.*
The blockchain was not reset but i was forging all the block and so maybe i'm on my own fork.

Thank you User1397367406, I'm glad that you didn't reset your main forging node. I agree that better you stop forging until other forger reach you. I was on that situation before, forging alone ahead of the rest, and my node resynchronized by itself when others reached my block height:

Hey!, I am synchronized again

I didn't stop my node, just I stopped to forge and it synchronized after 5.5 hours, here is the block history of that time (GMT+1):
...
260632 18/03/2015 23:00:54 0 0 0 NAS-5HZF-TPQ5-GSHF-7GV33 0 B 8708 %
260631 18/03/2015 22:59:17 0 0 0 NAS-5HZF-TPQ5-GSHF-7GV33 0 B 5386 %
260630 18/03/2015 16:29:59 0 0 0 NAS-H5AW-7J9F-7VCY-DJ8NB 0 B 2693 %
260629 18/03/2015 16:28:10 0 0 0 NAS-H5AW-7J9F-7VCY-DJ8NB 0 B 1482 %
...

[Adamlm]

Can someone post a link to the current working version of NAS ? This one which synchronizes. Thank you

[tiker]

Can someone post a link to the current working version of NAS ? This one which synchronizes. Thank you

At the moment, the network is under an attack by lcharles123.

Any public node will get you up to block 277934.  After that block you'll see the blocks appear and disappear as the network fights with lcharles123.

There is a plan to fix things but it will take some time.

The actual client is still in the first post of this thread - 1.1.3N1.

You'll need to modify the config to add the public nodes.  There's a post about that a few pages back.

[patapato]

We keep trying to get the network in sync. Be patient. Now there are forks and we are very few forgers. But we was in worst situation before February, so keep waiting, Nas network is still weak but not dead, just ill ;-)

[lcharles123]

I just let forging the patapato lease.

My node is lost in own fork, no conections... restarting download of blockchain.
How limit the java cache? My server does not support the war of forks.

Would not divide the coins of the current fork equally among all the nodes to avoid war forks?

[patapato]

How limit the java cache? My server does not support the war of forks.

  Funny man you, lcharles123  . Go out from the dark side and join us, the "good guys", we are going to win to the "villain" soon , and the Nas network will work healthy again.

Not bad at all to have this "stress test", as it is being useful to join forces, to understand weakness, failures and even bugs, to strengthen the Nas network, and to have now a core of new developers for the new life of Nas, with different views and criteria than the original ones.

To everybody, wait few time, the network nodes are going to be synchronized again to the stronger and longer blockchain fork.

Lesson learned, leasing balance for forging is not a good feature for PoS, we recommend to never lease again. Maybe we take out that feature from future versions of Nas. The trusting power of PoS is that an attacker needs to put his own stake on risk, but with leased stake, an attacker has nothing to risk, nothing to lose. Besides, the lease feature doesn't encourage more nodes on the network, on the contrary, it encourage the concentration of forging nodes, with the same potential problem as Bitcoin mining concentrated in few big pools.

[lcharles123]

How limit the java cache? My server does not support the war of forks.

  Funny man you, lcharles123  . Go out from the dark side and join us, the "good guys", we are going to win to the "villain" soon , and the Nas network will work healthy again.

Not bad at all to have this "stress test", as it is being useful to join forces, to understand weakness, failures and even bugs, to strengthen the Nas network, and to have now a core of new developers for the new life of Nas, with different views and criteria than the original ones.

To everybody, wait few time, the network nodes are going to be synchronized again to the stronger and longer blockchain fork.

Lesson learned, leasing balance for forging is not a good feature for PoS, we recommend to never lease again. Maybe we take out that feature from future versions of Nas. The trusting power of PoS is that an attacker needs to put his own stake on risk, but with leased stake, an attacker has nothing to risk, nothing to lose. Besides, the lease feature doesn't encourage more nodes on the network, on the contrary, it encourage the concentration of forging nodes, with the same potential problem as Bitcoin mining concentrated in few big pools.

Beautiful words.

Then the rent balance may well set precedents for an attack, it seems that the NEM, despite having the remote harvest, the score delegated is separate from node account, and coins found go to respectives account. It is a good substitute to lease balance, but do not know if it is implementable in NXT clone.

The rental period ends in 80 blocks

EDIT: the lease end, you can unblacklist my IP and redownload blocks.

[patapato]

How limit the java cache? My server does not support the war of forks.

  Funny man you, lcharles123  . Go out from the dark side and join us, the "good guys", we are going to win to the "villain" soon , and the Nas network will work healthy again.

Not bad at all to have this "stress test", as it is being useful to join forces, to understand weakness, failures and even bugs, to strengthen the Nas network, and to have now a core of new developers for the new life of Nas, with different views and criteria than the original ones.

To everybody, wait few time, the network nodes are going to be synchronized again to the stronger and longer blockchain fork.

Lesson learned, leasing balance for forging is not a good feature for PoS, we recommend to never lease again. Maybe we take out that feature from future versions of Nas. The trusting power of PoS is that an attacker needs to put his own stake on risk, but with leased stake, an attacker has nothing to risk, nothing to lose. Besides, the lease feature doesn't encourage more nodes on the network, on the contrary, it encourage the concentration of forging nodes, with the same potential problem as Bitcoin mining concentrated in few big pools.

Beautiful words.

You are welcome

Then the rent balance may well set precedents for an attack, ...

It also set precedents for a defense. I will never rent again, and I will try to take renting out of the Nas software.

The rental period ends in 80 blocks

EDIT: the lease end, you can unblacklist my IP and redownload blocks.

In fact, you are not blacklisted long ago, as it was unsuccessful. You earned a good fee for your work of testing the network in a war of forks. Keep it while Nas continue resurrecting, maybe you will buy a new server with it soon

Would not divide the coins of the current fork equally among all the nodes to avoid war forks?

The bigger is the forger, the more trusted he is (because he risks more). But I will partially listen to you, I don't want to forge alone.

[patapato]

NAS blockchain is riding again!! 

Forging was stable and healthy for 2 days yet, one main fork with several forgers (at least 4). Some transactions were made without problems, with confirmations from other forgers.

I like the old UI to see peers, the peer information is more complete and more clear in the old UI, so I put in my conf/nas.properties:

nas.enableUIServer=true
nas.uiServerPort=7872

and I connect to http://localhost:7872. Next image is what I see now, updating real time, very resdponsive UI. You can take the from here the peers that you want to put on nas.wellKnownPeers, in order that you get your client synchronized with the blockchain.



Compare it with the poorer information on the standard UI (few minutes before):



@tiker or @User1397367406, maybe you want to tell everybody something about how we got the blockchain to survive to the recent attack?

[michaelsss]

thank you for your working.

[lcharles123]

In fact, you are not blacklisted long ago, as it was unsuccessful. You earned a good fee for your work of testing the network in a war of forks. Keep it while Nas continue resurrecting, maybe you will buy a new server with it soon

Well, now it is stable. NXT was running together, but I'm not applying for it.

[tiker]

@tiker or @User1397367406, maybe you want to tell everybody something about how we got the blockchain to survive to the recent attack?

Well, we stopped the planet from spinning, then reversed the spinning direction to go back in time... wait, that was a Superman movie...

We found that the forked nodes agreed on all of the blocks up to a certain block height which was 277934.
We shut down a node once it deleted the forked blocks back to 277934.
We shut down a second node and deleted the block chain db.

The first node was brought up on an isolated network in a way that peers couldn't connect to it and it wouldn't attempt to connect out to anyone.
The second node was started up with a similar configuration but altered so that it would connect to the first isolated node.
Once the second node had downloaded the block chain from the first node (to verify it would be accepted by a new node) the two nodes started forging on their own disconnected from network under attack.
Transactions that were sent on the network under attack were not confirmed because of the attack.
Those transactions were copied and re-broadcasted on the isolated network and confirmed.  (The ones that were noticed at least with 1440 minute deadlines.)
After a few days, the other NAS forgers (that joined the XMPP chat room posted earlier) compared block chains from the different forked nodes and the isolated node.
After some testing with the different forks, seeing what worked and didn't, checking account balances, transactions sent during the attack, etc. it was decided that the fork running on the isolated network was the best NAS fork to resume working with.
The other forgers modified their clients to join the isolated network hidden from the public network and nodes under attack and ran various tests (forging, transactions, etc.) to verify everything worked.
Once we determined that the isolated network was at a point safe enough to re-introduce to the rest of the public and safe from the attack, we shutdown the nodes together, reconfigured them and started them up together for the rest of the world to access.

That brings us to today.
The NAS forks that were created from the attack will report that forging stopped a few days ago.  This is because the account which caused the problems was doing so with a balance lease that ended at a certain block height.  None of the forging accounts are forging on that chain due to transactions not being confirmed, possible sync issues for new NAS nodes, etc.

Anyone who finds their node stuck at a point where it is not forging or accepting new blocks is on a fork which will not be resolved by the client automatically.  If you are on one of these forks you can either restore a copy of your block chain from a backup taken before height 277934 (Apr. 5th ~14:00 GMT) or delete your node's database file and start over clean.

There is a public node "funkyvps3.funkymonkey.org" which can be added to the "wellKnownPeers" configuration option to get started.  This node is on block chain fork that the other NAS forgers agreed on.  This node also has the GUI interface open to the public at http://funkyvps3.funkymonkey.org:7873 for anyone who wants to login and check things.  This would be a good way to see if your local NAS nodes are on the same fork as the fork the rest of us agreed on.

Hopefully this sort of "attack" won't happen again.  We have learned a few things from this experience.

[User1397367406]

@tiker or @User1397367406, maybe you want to tell everybody something about how we got the blockchain to survive to the recent attack?

Oops! … I Did It Again 

No, seriously, it was really great teamwork from us three.
I'm glad we did it.

NAS is back on track again 

My main node is connected and forging on block chain fork that the other NAS forgers agreed on.

[User1397367406 make a small wave with his hand] You don't need to see my second node.
This is not the node you're looking for.
...
Wait that's another movie

I'm currently trying to sync a second node from scratch in order to confirm the fork is legit.
Seems to sync all right: block height 87443 and still downloading.
I've put our three node in "nas.wellKnownPeers".
I'll keep you informed.