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

[zvs]

what the devil

2012-09-06 06:53:54.077480 GOT BLOCK FROM MINER! Passing to bitcoind! http://blockexplorer.com/block/000000000000013fcf0d071e54c00838654cc2328b463bea33894ed51eec96dd
2012-09-06 06:53:54.101687 GOT SHARE!  1eec96dd prev 699c48c8 age 8.58s DEAD ON ARRIVAL

http://nogleg.com:9332/static/share.html#000000000000013fcf0d071e54c00838654cc2328b463bea33894ed51eec96dd

ok, so i knew it did it with orphans, but 'dead on arrival'?

(ed: hmm, no, that's not it...  bizarre)

it looks like i trumped myself by it being passed to bitcoind and acknowledged there first?

[1713894091]

1713894091

[1713894091]

1713894091

[rav3n_pl]

what the devil

2012-09-06 06:53:54.077480 GOT BLOCK FROM MINER! Passing to bitcoind! http://blockexplorer.com/block/000000000000013fcf0d071e54c00838654cc2328b463bea33894ed51eec96dd
2012-09-06 06:53:54.101687 GOT SHARE!  1eec96dd prev 699c48c8 age 8.58s DEAD ON ARRIVAL

http://nogleg.com:9332/static/share.html#000000000000013fcf0d071e54c00838654cc2328b463bea33894ed51eec96dd

ok, so i knew it did it with orphans, but 'dead on arrival'?

(ed: hmm, no, that's not it...  bizarre)

it looks like i trumped myself by it being passed to bitcoind and acknowledged there first?

It IS possible
P2pool is checking every share because every can BE a block solution, but not every share can fit into p2pool share chain because of fast long pooling.
Thats why it is crucial to miner software send stale shares Sometimes orphaned/doa share CAN be a bitcoin block
This way you only lost one pool share, but get block

[kano]

... or as I've stated before ...
Only transfer the block header, coinbase txn and merkle tree by default then get the recipient to say if it wants the transactions and which ones (a list) from the merkle tree.
Most bitcoinds already have ALL txn but the coinbase, so that is a massive waste and always has been.
Secondly, why are the txn's re-validated, shouldn't they have been validated when they were received the first time?

I'm certainly by far not an expert at this.. but I would assume that you there is a "trust no one" factor here, you can't safely assuming incoming transactions are validated already.

He meant the clients might have already seen the transaction broadcast before, so would have validated it themselves already. That is indeed already cached, but there are a lot of transactions that don't get seen by any given node (about 10% IIRC), and the back-and-forth of requesting missing transactions would add delays - though it might make sense on top of the basic block distribution parallelization I mentioned, but it will need to be benchmarked.

Well the request would simply be a full request (from the merkle tree) of the missing transactions.
A single request not multiple.
Very simply to know what is missing since the merkle tree lists all the txns and thus the reply would be those at the bottom of the tree you don't already have.

As for 10%? Where do you get that figure from?

I wrote a little transaction processing module to add to bitcoind (that I run all the time), that, among other things, specifically tells me the number of missing transactions.
After about a day from a restart, it was EXTREMELY rare for it to not have every txn except of course the coinbase (which no one will already have except the person/pool who found the block - and why of course I said that should always be sent)

I'd very much doubt that 10% number.
However, even if it was as extremely high as 10%, it would be way faster to send back a reply of 10% of the transaction numbers, then get those 10%, than sending the full 100% every time when 90% of them are unnecessary.

As for the memory pool txn's - why would there be a txn in there that wasn't already verified?
What extra verification is done when the txn's are received with a new block that isn't already done as each txn comes in?

[zvs]

... or as I've stated before ...
Only transfer the block header, coinbase txn and merkle tree by default then get the recipient to say if it wants the transactions and which ones (a list) from the merkle tree.
Most bitcoinds already have ALL txn but the coinbase, so that is a massive waste and always has been.
Secondly, why are the txn's re-validated, shouldn't they have been validated when they were received the first time?

I'm certainly by far not an expert at this.. but I would assume that you there is a "trust no one" factor here, you can't safely assuming incoming transactions are validated already.

He meant the clients might have already seen the transaction broadcast before, so would have validated it themselves already. That is indeed already cached, but there are a lot of transactions that don't get seen by any given node (about 10% IIRC), and the back-and-forth of requesting missing transactions would add delays - though it might make sense on top of the basic block distribution parallelization I mentioned, but it will need to be benchmarked.

Well the request would simply be a full request (from the merkle tree) of the missing transactions.
A single request not multiple.
Very simply to know what is missing since the merkle tree lists all the txns and thus the reply would be those at the bottom of the tree you don't already have.

As for 10%? Where do you get that figure from?

I wrote a little transaction processing module to add to bitcoind (that I run all the time), that, among other things, specifically tells me the number of missing transactions.
After about a day from a restart, it was EXTREMELY rare for it to not have every txn except of course the coinbase (which no one will already have except the person/pool who found the block - and why of course I said that should always be sent)

I'd very much doubt that 10% number.
However, even if it was as extremely high as 10%, it would be way faster to send back a reply of 10% of the transaction numbers, then get those 10%, than sending the full 100% every time when 90% of them are unnecessary.

As for the memory pool txn's - why would there be a txn in there that wasn't already verified?
What extra verification is done when the txn's are received with a new block that isn't already done as each txn comes in?

well, you probably have more connections than the avg bitcoin user.

many will only have 8

[kano]

...
well, you probably have more connections than the avg bitcoin user.

many will only have 8

9 on my main one and 8 on my secondary one

[BR0KK]

im trying p2pool now

I have sucessfuly installed it on my win 7 machine and im pointing 200MH/s to it.


How do the statistics work? They are not displaying anything atm? (Does this need time?)

I want to use P2Pool, bitcoind and my miner on three separate machines. HOw can i get this to work?

Code:

P2Pool: 17433 shares in chain (17222 verified/17437 total) Peers: 10 (0 incoming)
 Local: 250MH/s in last 10.0 minutes Local dead on arrival: ~0.0% (0-10%) Expected time to share: 2.9 hours
 Shares: 0 (0 orphan, 0 dead) Stale rate: ??? Efficiency: ??? Current payout: 0.0000 BTC
 Pool: 277GH/s Stale rate: 6.7% Expected time to block: 11.6 hours
New work for worker! Difficulty: 0.999985 Share difficulty: 591.506619 Total block value: 50.511625 BTC including 622 transactions

What does this mean? is it working properly?


Thx for hlp

[tiberiandusk]

Looks like it is working correctly. Go to http://127.0.0.1:9332 in your browser to see some more info. You can run P2Pool on one computer and direct your miners to that computer's IP address instead of the localhost address.

[mdude77]

im trying p2pool now

I have sucessfuly installed it on my win 7 machine and im pointing 200MH/s to it.


How do the statistics work? They are not displaying anything atm? (Does this need time?)

I want to use P2Pool, bitcoind and my miner on three separate machines. HOw can i get this to work?

Code:

P2Pool: 17433 shares in chain (17222 verified/17437 total) Peers: 10 (0 incoming)
 Local: 250MH/s in last 10.0 minutes Local dead on arrival: ~0.0% (0-10%) Expected time to share: 2.9 hours
 Shares: 0 (0 orphan, 0 dead) Stale rate: ??? Efficiency: ??? Current payout: 0.0000 BTC
 Pool: 277GH/s Stale rate: 6.7% Expected time to block: 11.6 hours
New work for worker! Difficulty: 0.999985 Share difficulty: 591.506619 Total block value: 50.511625 BTC including 622 transactions

What does this mean? is it working properly?


Thx for hlp

With that small hash rate, you might want to check out p2pmining.com.  It seems to work better for smaller miners, and you get merged mining.

M

[BR0KK]

i have more to point towards p2pool--> about 4 GHs!

Wanted to try first, to understand how it works

[rav3n_pl]

i have more to point towards p2pool--> about 4 GHs!

Wanted to try first, to understand how it works

Yes, it is working correctly.
On current pool rate you should get share about every 3hrs (using 250MH power).
If you want point more miners to your pool on local network just point miners to http://ip.of.your.node:9332, just be sure that you have port 9332 open.
Also try open port 9333 for incoming connection to get faster info about p2pool shares and new bitcoin blocks.

[doubleganger]

I have two jalapenos on pre-order and was wondering if it would be a waste of time to go ahead and try to get set up using cpu mining just to have the set up partly done when the jalapenos get here (if they ever do, that is)? For one thing, I wouldn't want to do anything that would negatively impact the pool. I realize I probably wouldn't get ANY bitcoin from it. Please excuse me if this has already been asked.

[mdude77]

I have two jalapenos on pre-order and was wondering if it would be a waste of time to go ahead and try to get set up using cpu mining just to have the set up partly done when the jalapenos get here (if they ever do, that is)? For one thing, I wouldn't want to do anything that would negatively impact the pool. I realize I probably wouldn't get ANY bitcoin from it. Please excuse me if this has already been asked.

I would not use CPU mining.

M

[doubleganger]

Thanks.  The setup would probably be different enough that it wouldn't gain  me anything anyway.

[racerguy]

Cpu mining would be fine to make sure everything's running fine before your asics arrive - as long as you don't mind waiting a few days to get a single share.

[BR0KK]

I've pointed 400MHs at P2Pool atm; First payout happened today.


Im quite happy with it (looked harder to install than it is). I might switch completely to this solution in a while!

Now i need to separate the bitcoind installation from my p2 pool server.
So that i have three different machines running (i like them separate! debuging gets easier)

I have a spare Raspberry pi (that is no Use as a miner with my Ztex boards). Would it be possible to run one of the two software on that machine? (how much CPU power does P2Pool need? i pointed a one core and 1.5gib of ram towards this VM. Seems enough....)

I need to forward the pools ports with my fritzbox this evening.



Is there a way to dispplay stst to the internet?
I use BTCmon with eclipse and my other pools. Is there something similar for P2Pool?

[Krak]

Is there a way to dispplay stst to the internet?
I use BTCmon with eclipse and my other pools. Is there something similar for P2Pool?

Just forward 9332 to the machine running p2pool and go to http://your.ip:9332 to see stats. Or get a domain name for your network if you have a dynamic IP address. You can see an example of this at http://p2pmining.com:9332

[doubleganger]

Cpu mining would be fine to make sure everything's running fine before your asics arrive - as long as you don't mind waiting a few days to get a single share.

Thanks for the input.  No I wouldn't expect to get any  bitcoin doing it.

[dellech]

Block 198040 ... another private fork of p2pool ?

[mdude77]

Block 198040 ... another private fork of p2pool ?

That would explain it...

M

[twmz]

Block 198040 ... another private fork of p2pool ?

That would explain it...

M

Damn.  I guess I'll have to blacklist the addresses in those blocks because this is getting annoying.