[1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool

[baloo_kiev]

i just turned my stuff back to p2pool for a final hurrah i suppose you could say, before i shut everything down

but doesn't anyone find it concerning that someone has 33% of the hashrate?  i also was unable to locate the IP address of this person, so they're probably behind a firewall, without many outgoing connections, judging from the other nodes I checked (IP address reported as relaying their share is quite a spread)

just found it curious since i've gotten 2 orphans out of 6 already,  both because of the double share from 1Nasty

although i guess it could be partally attributed to using a US server

Why would it be disconcerting?  p2pool is really small.  there are filthy rich miners who have more than the entire hash rate of p2pool.

M

because it decreases efficiency of other people using p2pool when someone is using inefficient connection and is able to put out 33% of hashrate?

does it need further explaining?

Yes please! How a high-latency miner with 33% total hashrate can decrease other's efficiency?

[rav3n_pl]

Yes please! How a high-latency miner with 33% total hashrate can decrease other's efficiency?

Easy. He just need found 2 shares in row faster than rest of nodes.
share chain:
S1
S2<S1
...
Sa<S9
Sb<Sa - yours
Sc<Sa - thiers
Sd<Sc - thiers n2 -> your Sb is orphaned
Se<Sd - someone else
Sf<Sd - thier n3
Sg<Sf - thier n4 -> Se is orphaned too
He got 4 shares, 2 other orpahned in proecess. Because of high latency he is not aware of other shares.

[baloo_kiev]

Yes please! How a high-latency miner with 33% total hashrate can decrease other's efficiency?

Easy. He just need found 2 shares in row faster than rest of nodes.
share chain:
S1
S2<S1
...
Sa<S9
Sb<Sa - yours
Sc<Sa - thiers
Sd<Sc - thiers n2 -> your Sb is orphaned
Se<Sd - someone else
Sf<Sd - thier n3
Sg<Sf - thier n4 -> Se is orphaned too
He got 4 shares, 2 other orpahned in proecess. Because of high latency he is not aware of other shares.

But his probability to find a share faster than others is only 33%, and about 11% for finding two shares before the rest. If he fails, his shares will be orphaned. As a result, he harms himself more, just as planned!

[rav3n_pl]

Maybe he is hitting full power for 10 sec every 20 sec? ;]
Joking, but it is all about luck. Higher latency means that his shares will be "punished" each block.
How fast he can create 2 shares in row depends on his power and how good he is connected to network.

[freshzive]

getting close to 1thash!

[twobits]

So what do we want him to do?   Connect to more nodes?   

[gyverlb]

So what do we want him to do?   Connect to more nodes?  

Nothing. As long as he doesn't have more than 50% of the P2Pool hashrate he is giving us free coins. If he doesn't want to, he should read the guide in my signature.

[GrapeApe]

Are we supposed to be using the newshare branch from forresv's github or the main one?

[Boing7898]

in command line:
ulimit -n 8192
and run p2pool

and check later if errors show.

Open files in linux-like system is also opened connections.

I edited /etc/security/limits.conf with limits of 10000 and I've been running p2pool for nearly 3 hours without a crash.
Weird though, because before I upgraded p2pool it was working, even with the limit set at 1024. Maybe it's because more users joined my pool?
Who knows. 

Code:

$ lsof -c python | wc -l

On the node will tell you how many files p2pool has opened.

1494 files, wow.
That's why it was crashing, the limit was at 1024 before.
Thanks for the tip!

Can you do instead, replacing P2POOL_PID with run_p2pool.py's PID,

Code:

$ lsof -p P2POOL_PID

and pastebin the output and send it to me? P2Pool shouldn't be using that many files - seeing which it has open could help find the issue.

Hey, did you get my private message?

[forrestv]

Are we supposed to be using the newshare branch from forresv's github or the main one?

Master for now. newshare will be merged in and released very soon.

Hey, did you get my private message?

Yes, I did. It looked like all the open files were sockets from miners, so there may not be anything you can do. Do you expect to have hundreds of simultaneous connections from miners? If so, you might want to look at load-balancing over multiple P2Pool instances. If not, someone may be attacking your node by making hundreds of connections to it.

[Boing7898]

Are we supposed to be using the newshare branch from forresv's github or the main one?

Master for now. newshare will be merged in and released very soon.

Hey, did you get my private message?

Yes, I did. It looked like all the open files were sockets from miners, so there may not be anything you can do. Do you expect to have hundreds of simultaneous connections from miners? If so, you might want to look at load-balancing over multiple P2Pool instances. If not, someone may be attacking your node by making hundreds of connections to it.

Hmm. Can you give me more info about load-balancing? The only things I found about it are load-balancing the network connection, not 2 instances of a process.
Also, is it normal that the pool is finding valid shares but Payout if a block were found NOW is stuck at 0, even though fee is set to 2%?

[bitpop]

Load balance just like an httpd

[forrestv]

Are we supposed to be using the newshare branch from forresv's github or the main one?

Master for now. newshare will be merged in and released very soon.

Hey, did you get my private message?

Yes, I did. It looked like all the open files were sockets from miners, so there may not be anything you can do. Do you expect to have hundreds of simultaneous connections from miners? If so, you might want to look at load-balancing over multiple P2Pool instances. If not, someone may be attacking your node by making hundreds of connections to it.

Hmm. Can you give me more info about load-balancing? The only things I found about it are load-balancing the network connection, not 2 instances of a process.
Also, is it normal that the pool is finding valid shares but Payout if a block were found NOW is stuck at 0, even though fee is set to 2%?

Use something like the "balance" program to do TCP load balancing between multiple P2Pool instances. You'll have to start two separate P2Pool instances listening on different worker ports, then use balance to make a public port that round-robins between them.

The fee is probabilistic - 2% of shares will go to you, not 2% of every share, so you're probably just unlucky at the moment.

[forrestv]

P2Pool release 13.0  - HARDFORK, UPGRADE REQUIRED - commit hash: f3a0e8dfcd872716123771db7900cdcd963b91ce

Windows binary: http://u.forre.st/u/xqerwrpk/p2pool_win32_13.0.zip
Windows binary signature: http://u.forre.st/u/viejmrru/p2pool_win32_13.0.zip.sig
Source zipball: https://github.com/forrestv/p2pool/zipball/13.0
Source tarball: https://github.com/forrestv/p2pool/tarball/13.0

After 50% of each P2Pool's mining power has upgraded, warnings will be displayed to everyone who hasn't upgraded. Approximately 24 hours after 95% of the mining power has upgraded, the switch will happen.

Changes:
* Hardfork at 95% upgraded:
** Bitcoin share period increased from 10 to 30 seconds to cater to ASIC miners. Avalon/BFL/ASICMINER devices should start working well after this.
** Litecoin share period increased from 10 to 15 seconds
** Litecoin payouts spread over 3 block-lengths instead of 12, reducing dust payouts
** Transaction pre-forwarding greatly simplified, allowing future network traffic reductions
** Maximum share difficulty multiplier increased from 10x to 30x to give more freedom to below share difficulty adjustments
** OP_RETURN used in last txout to prevent UTXO database spam
** Stratum nonce length increased from 4 to 8 bytes, allowing for future Avalon support without having to use the "avalon" branch

* Automatically increase share difficulty to prevent payouts below "dust threshold", 0.001 BTC and 0.03 LTC
* Automatically increase share difficulty to prevent any single node from making more than 5% of shares, by default
* Worker username parameters (+PSEUDOSHARE_DIFF/SHARE_DIFF) not longer have to be in a specific order
* Support for submitblock RPC call in new Litecoin versions
* Fixed incompatibility with ASICMINER BE Blade
* Updated bootstrap address list

[CartmanSPC]

Thanks for all the hard work! Will update my nodes tomorrow.

[Boing7898]

Updated my node right now, looks fine. Thanks! 

[daemondazz]

cryptominer.org updated now as well, running BTC, LTC and TRC.

[yuyi]

http://mine.yuyi.tw:9332/ updated.

looks great.

[centove]

http://ask.gxsnmp.org:9332/ updated.

[baloo_kiev]

It's interesting how AM blades wil perform after the switch. According to my calculations, they must still have about 18% DOA