p2pool - Decentralized, Absolutely DoS-Proof, Pool Hopping-Proof Pool [archival]

[bz]

I'm ignorant about p2pool but thinking about trying it out. I'd like to be able to see some stats (at least current p2pool hash rate, graphs over time would be great). Are such stats visible to outsiders anywhere?

I'd also like to know if I should expect more stales and if so, how many more are typical?

[broken]

I'm ignorant about p2pool but thinking about trying it out. I'd like to be able to see some stats (at least current p2pool hash rate, graphs over time would be great). Are such stats visible to outsiders anywhere?

I'd also like to know if I should expect more stales and if so, how many more are typical?

here is a graph:
http://u.forre.st/p2pool/600.png

I have 10% stales, which is pretty high

[forrestv]

So stales are high ,righ?

I'm ignorant about p2pool but thinking about trying it out. I'd like to be able to see some stats (at least current p2pool hash rate, graphs over time would be great). Are such stats visible to outsiders anywhere?

I'd also like to know if I should expect more stales and if so, how many more are typical?

P2Pool stales are different from normal stales - they still contribute to the pool's useful hashrate (its ability to solve blocks), but you don't get credit within the payout counting system from them. However, everyone gets about the same stale rate - right now it's 25% - so it's fair, because the would-be-losses are evenly distributed, and so have no effect.

The P2Pool client displays an "efficiency" percentage based on your stale proportion relative to everyone else's. Normally, it should be around 100%, which means that you have average stales, but it might be lower if your setup is bad.

[BTCurious]

P2Pool stales are different from normal stales - they still contribute to the pool's useful hashrate (its ability to solve blocks), but you don't get credit within the payout counting system from them. However, everyone gets about the same stale rate - right now it's 25% - so it's fair, because the would-be-losses are evenly distributed, and so have no effect.

Thanks for explaining. I was initially deterred by this supposedly high stale rate.

[forrestv]

Thanks for explaining. I was initially deterred by this supposedly high stale rate.

Yeah, I've thought about renaming P2Pool-stales to something else, because they can't be compared to normal-pool-stales (and discourage people). I haven't though of a good name, though... Any ideas?

[BTCurious]

Yeah, I've thought about renaming P2Pool-stales to something else, because they can't be compared to normal-pool-stales (and discourage people). I haven't though of a good name, though... Any ideas?

Do you need to name them at all? They're rather meaningless… I guess the client shows them though?

[forrestv]

Do you need to name them at all? They're rather meaningless… I guess the client shows them though?

The client shows them and the numbers are useful for checking whether everything's working correctly.

Code:

22:18:09.886410 Pool: 5279MH/s in 8694 shares (8698/8698 verified) Recent: 24.11% >1272MH/s Shares: 2 (0 orphan, 0 dead) Peers: 1
22:18:09.886509 Median stale proportion: 0.236220472441
22:18:09.886543     Own: 0.0
22:18:09.886571     Own efficiency: 130.93%

(This is only based on the two shares, so it's not accurate. It will converge to around 100%)

[bz]

@broken, thanks for the link. I was anticipating more were using the pool. I really like the idea of p2pool, seems like others should too. Especially after recent attacks on pools. Hopefully the hash rate will go up so blocks can be found faster. Or, maybe difficulty will continue to go down and make solving blocks easier...

@forrestv, thanks for explaining the stales. I'd would probably leave if I saw them without that explanation. I'll take one of my machines and see how it goes, perhaps next weekend.

[Mike Hearn]

Maybe just show the "efficiency" score without any mention of stales. It's bad to have people leave purely because of a terminological issue. A flag could let you dump them if you need to know the exact number. If you absolutely insist on showing them try "pool-accounted blocks"

[bronan]

Its a nice concept but yes setting it up is some work in progress
For now even though it costs some time and effort when you finally got it running, it really runs well as long as it does not loose connection with the net.
Now a gui would be nice and most important a way to get stats from each worker/client and so on.
Cause we all thrive on the numbers and stats, without them any miner feels naked and lonely
So to make a long story short STATS, STATS and more STATS

[forrestv]

P2Pool 0.7.2 - tag: release-0.7.2

EDIT: Found a critical bug with sharechain ranking, released 0.7.1
EDIT: Found a bug with litecoin mining with Artforz's cpuminer, released 0.7.2

Windows binary: http://u.forre.st/u/pugaormc/p2pool_0.7.2_55673c5.zip

Changes:
* Memory usage decreased by about 40% by only caching the last 1000 block headers instead of all
* Startup time greatly decreased due to previous change and better share caching
* Peers are judged using a heuristic, so the connection count goes to the maximum in seconds instead of many minutes
* Fixed bug in judging which of disjoint sharechains should be mined on
* Some efficiency improvements for miners
* Fixed "KeyError" bug
* Stale and efficiency displays now have a confidence interval:

15:08:47.435707 Pool stales: 12% Own: 11±4% Own efficiency: 101±4%

* All cache files (addrs.dat, shares.*, headers) are now created in data/ subdirectory
* addrs.dat no longer uses sqlite, instead using plain text, which should be more robust

Everyone should upgrade as soon as possible, as we have several competing sharechains due to that bug!

Phoenix-miner support is currently broken, but stock cgminer works very well.

[LightRider]

Minor cosmetic problem, the plus/minus symbol displays incorrectly on my machine:

Code:

Own efficiency: 71┬▒47%

Win7x64

[iongchun]

Hi forrestv, I use Bitcoin address as my miners' username, but I guess merged-mined namecoins are paid to a address of connected namecoind.
Could I specify a Namecoin address in miners' username too?

[LightRider]

What does "No type for share1as" mean?

[forrestv]

Hi forrestv, I use Bitcoin address as my miners' username, but I guess merged-mined namecoins are paid to a address of connected namecoind.
Could I specify a Namecoin address in miners' username too?

No, there is no way to affect the Namecoin payouts. P2Pool is just blindly requesting work from namecoind.

What does "No type for share1as" mean?

It means that you're connected to somebody running an outdated client and they're trying to send you old-style shares. I probably should have made this a debugging-mode-only message, because I expected it to appear.. I think I'll do that now

[broken]

From Git Commit 66cbd54ce70f7135dedd6e3799f0840394ab9cfb:

  1 satoshi is always donated so that a list of p2pool generated blocks can be easily found by looking at the donation address

Can you provide the bitcoin-address or blockexplorer link please.

[finway]

I've got this error, need help  


And i think moderator should make an exception for p2pool, make it sticked,
cause it's good for decentraliztion.

[forrestv]

From Git Commit 66cbd54ce70f7135dedd6e3799f0840394ab9cfb:

  1 satoshi is always donated so that a list of p2pool generated blocks can be easily found by looking at the donation address

Can you provide the bitcoin-address or blockexplorer link please.

http://blockexplorer.com/address/1Kz5QaUPDtKrj5SqW5tFkn7WZh8LmQaQi4 for Bitcoin. Here's how the rest can be computed:

Code:

>>> SCRIPT = '4104ffd03de44a6e11b9917f3a29f9443283d9871c9d743ef30d5eddcd37094b64d1b3d8090496b53256786bf5c82932ec23c3b74d9f05a6f95a8b5529352656664bac'.decode('hex')
>>> from p2pool.bitcoin import data, networks
>>> data.script2_to_human(SCRIPT, networks.BitcoinMainnet)
'Pubkey. Address: 1Kz5QaUPDtKrj5SqW5tFkn7WZh8LmQaQi4'
>>> data.script2_to_human(SCRIPT, networks.NamecoinMainnet)
'Pubkey. Address: NFZScDyN9GRRFchLmuCpyJGRHvXPprfABB'
>>> data.script2_to_human(SCRIPT, networks.LitecoinMainnet)
'Pubkey. Address: LeD2fnnDJYZuyt8zgDsZ2oBGmuVcxGKCLd'

[forrestv]

I've got this error, need help  

I've never seen that error before.. Are you using wallet encryption? If so, it's probably because Bitcoin can't generate any more keys without the passphrase, and you should do as the error says - run bitcoind.exe keypoolrefill.

And i think moderator should make an exception for p2pool, make it sticked,
cause it's good for decentraliztion.

It would be nice if they would do that. However, we can keep it up there by keeping it active.. and once it is one of the top ten pools it will be stickied anyway.

[finway]

I've never seen that error before.. Are you using wallet encryption? If so, it's probably because Bitcoin can't generate any more keys without the passphrase, and you should do as the error says - run bitcoind.exe keypoolrefill.

yes, i'll try that when i got home.

And, can you make this pool running SMPPS mode , just like Eligius?

I think that'll attract more miner, avoiding the "Chicken & Egg" thing.

EDIT:

I've got another exception:


EDIT again:
Ok, i switched to bitcoin 0.5, it seems OK now!
bu still get some "error like" messages: