Pages:
Author

Topic: p2pool - Decentralized, Absolutely DoS-Proof, Pool Hopping-Proof Pool [archival] - page 13. (Read 35481 times)

legendary
Activity: 1500
Merit: 1021
I advocate the Zeitgeist Movement & Venus Project.
P2Pool 0.8.1 - tag: release-0.8.1

Windows py2exe binary: http://u.forre.st/u/eyhowtzt/p2pool_0.8.1_c7feb00.zip
Source tarball: https://github.com/forrestv/p2pool/tarball/release-0.8.1

Changes:
* Pool stales are published and computed in a much more accurate way
* Merkle root caching isn't hurt by having miners with different usernames
* The HTTP method bug present with poclbm and ufasoft is fixed
* Some other efficiency and memory improvements for miners

You are awesome, thanks!
hero member
Activity: 516
Merit: 643
P2Pool 0.8.1 - tag: release-0.8.1

Windows py2exe binary: http://u.forre.st/u/eyhowtzt/p2pool_0.8.1_c7feb00.zip
Source tarball: https://github.com/forrestv/p2pool/tarball/release-0.8.1

Changes:
* Pool stales are published and computed in a much more accurate way
* Merkle root caching isn't hurt by having miners with different usernames
* The HTTP method bug present with poclbm and ufasoft is fixed
* Some other efficiency and memory improvements for miners
legendary
Activity: 1500
Merit: 1021
I advocate the Zeitgeist Movement & Venus Project.
Still getting that weird HEAD POST error when using ufasoft miner, and cgminer continues to crash my display drivers. Wish I could contribute to the pool.

Luckily, p2pool seems to play nice with bitcoind when bitcoinqt isn't running.
hero member
Activity: 896
Merit: 1000
p2pool is now cruising at 48GH/s and still rising. If this continues it may join the top 10 pools (it only needs to reach 100GH/s to do so).
legendary
Activity: 1500
Merit: 1021
I advocate the Zeitgeist Movement & Venus Project.
I don't know who's fault it is, but every time I connect p2pool to bitcoin-qt 5.1, the bitcoin client crashes.

Code:
Problem signature:
  Problem Event Name: APPCRASH
  Application Name: bitcoin-qt.exe
  Application Version: 0.0.0.0
  Application Timestamp: 4d44aa00
  Fault Module Name: bitcoin-qt.exe
  Fault Module Version: 0.0.0.0
  Fault Module Timestamp: 4d44aa00
  Exception Code: c0000005
  Exception Offset: 00c98d59
  OS Version: 6.1.7601.2.1.0.256.48
  Locale ID: 1033
  Additional Information 1: 0a9e
  Additional Information 2: 0a9e372d3b4ad19135b953a78882e789
  Additional Information 3: 0a9e
  Additional Information 4: 0a9e372d3b4ad19135b953a78882e789
hero member
Activity: 516
Merit: 643
How can i see if merged mining is working? P2pool sometimes logs:

 Got new merged mining work! Difficulty: 409454.724619

So I think it should be working, but namecoind shows no balance or transactions.
Any ideas?

As long as it shows that, everything is working. Keep in mind that you're mining solo, and it can be a long time between blocks. You can test namecoin block submission if you run with --debug - you'll get lots of lines saying "MERGED RESULT: False" or "MERGED RESULT: True" if you get a block. I know that at least gmaxwell has recently gotten a namecoin block.
hero member
Activity: 516
Merit: 643
Just curious : why does p2pool report a number of shares roughly covering 48 hours ? Is it to make it possible to switch to another memory period later or did I misinterpret something else ?

Currently, 48 hours of shares are held in order to make it possible to switch to 24 hours, which will happen in a few days.

It requires double the number of shares because an individual share depends on having a history to be verified. If the memory period is 24 hours, we keep the last 48 hours of shares, so we can verify the last 24 hours.
newbie
Activity: 55
Merit: 0

Some questions for fellow p2pool users : is anyone using cgminer with p2pool as primary pool and several other fallback pools ?

I ask because I see my cgminer-2.1.1 instances using fallback pools on occasions (on temporary network slowdowns maybe) and it seems they can get stuck for large periods using the fallback pool(s). At least my p2pool node logs exhibit large variations of hashing speed (between ridiculously low rates to near theoretical max hashrate) over large periods (the oscillations seem to span several hours).
All clients are cgminer 2.1.0 and 2.1.1 with similar configuration, I often see 1-difficulty share being submitted. It's probably a cgminer bug but I wonder if anyone saw this kind of log :
Code:
15:30:04.698403 Worker submitted share with hash > target:
15:30:04.698579 hash  : 474e164967bdceaedb355c9bfcb8f96fadcfc1ad25a2b00cfb148770
15:30:04.698713 target: 578ae00000000000000000000000000000000000000000000000000
Might be a share for a "classic" pool being submitted to p2pool in error (I think 2.0.8 had this kind of problem and believed it to be fixed in 2.1.0).
Anyone can confirm seeing this kind of behavior on cgminer (or another miner supporting fallbacks) ?

I´m using cgminer 2.0.8 with two backup pools. I noticed the same behavior, that the miners spend much time on the backup pools. I m now using only one backup pool and now only about 0,5% of the hashing power goes to the backup pool. I never saw that log of yours here, but sometimes I get "[2012-01-02 17:19:49] Pool 0 not providing work fast enough"


I have another question:
I´m not generating any namecoins.
 ~/.namecoin/bitcoin.conf has rpcuser, rpcpassword and rpcport set. Im running p2pool like this:

python run_p2pool.py bitcoinRpcUser bitcoinRpcPass  --merged-url http://127.0.0.1:8882/ --merged-userpass namecoinRpcUser:namecoinRpcPass

How can i see if merged mining is working? P2pool sometimes logs:

 Got new merged mining work! Difficulty: 409454.724619

So I think it should be working, but namecoind shows no balance or transactions.
Any ideas?
hero member
Activity: 896
Merit: 1000
I'm sorry for your loss, bronan... P2Pool has very high variance due to the long time between blocks and the short memory period, which is currently 12 hours, but will switch to 24 hours when the above change takes effect. This should make anyone who mines only during the day (or any other daily period) not be hurt by variance due to the memory period.
Just curious : why does p2pool report a number of shares roughly covering 48 hours ? Is it to make it possible to switch to another memory period later or did I misinterpret something else ?

Some questions for fellow p2pool users : is anyone using cgminer with p2pool as primary pool and several other fallback pools ?

I ask because I see my cgminer-2.1.1 instances using fallback pools on occasions (on temporary network slowdowns maybe) and it seems they can get stuck for large periods using the fallback pool(s). At least my p2pool node logs exhibit large variations of hashing speed (between ridiculously low rates to near theoretical max hashrate) over large periods (the oscillations seem to span several hours).
All clients are cgminer 2.1.0 and 2.1.1 with similar configuration, I often see 1-difficulty share being submitted. It's probably a cgminer bug but I wonder if anyone saw this kind of log :
Code:
15:30:04.698403 Worker submitted share with hash > target:
15:30:04.698579 hash  : 474e164967bdceaedb355c9bfcb8f96fadcfc1ad25a2b00cfb148770
15:30:04.698713 target: 578ae00000000000000000000000000000000000000000000000000
Might be a share for a "classic" pool being submitted to p2pool in error (I think 2.0.8 had this kind of problem and believed it to be fixed in 2.1.0).
Anyone can confirm seeing this kind of behavior on cgminer (or another miner supporting fallbacks) ?
hero member
Activity: 516
Merit: 643
General notice: All users must be using P2Pool version 0.7.1 or greater before Thu Jan 5 23:11:45 UTC (Unix time 1325805105) (Thu Jan 5 18:11:45 EST). P2Pool is switching from a 12 hour maximum payout period to a 24 hour maximum payout period.

but as i see it only those mining the last 24hours get paid.
This is the exepected behaviour. Although I believe you are paid for the last 48 hours at the time a block is found (at least the p2pool client remembers shares for this period from what I can read in its logs).

I'm sorry for your loss, bronan... P2Pool has very high variance due to the long time between blocks and the short memory period, which is currently 12 hours, but will switch to 24 hours when the above change takes effect. This should make anyone who mines only during the day (or any other daily period) not be hurt by variance due to the memory period.
hero member
Activity: 896
Merit: 1000
so i think for those who stop mining in between and game its a bad choice
« hop-proof » is self-explanatory :-)

but as i see it only those mining the last 24hours get paid.
This is the exepected behaviour. Although I believe you are paid for the last 48 hours at the time a block is found (at least the p2pool client remembers shares for this period from what I can read in its logs).
hero member
Activity: 774
Merit: 500
Lazy Lurker Reads Alot
well i think done enough test at p2pool and got not payed at all again.
so i think for those who stop mining in between and game its a bad choice
the shares i did all disappeared and got paid to who ever is the lucky one, but not to me.
but as i see it only those mining the last 24hours get paid.
hero member
Activity: 896
Merit: 1000
Works great. Even motivated me enough to take the time to watercool my 5970 with a faulty fan today and added it to my rig.
staff
Activity: 4172
Merit: 8419
At forestv's suggestion I decided to give p2pool another try—  Wow. It's really matured a lot.

I've had about 8GH/s on it for a couple of hours with no issues.  It was a cinch to setup, even with merged mining.

I think P2Pool makes mining fun again.
hero member
Activity: 896
Merit: 1000
Just a quick message to say that all my previous problems with p2pool have disappeared. It now fits nicely in RAM on my VPS, no more swapping, bitcoind-0.5.1 seems stable too (previous versions did crash regularly, maybe because of the out-of-memory killer), it runs without problems for more than a week now (didn't found a block yet though...).
legendary
Activity: 1500
Merit: 1021
I advocate the Zeitgeist Movement & Venus Project.
Only seems to appear when there is "new work for worker".
New version. Bitcoinqt 0.5.1 on Win7x64.

What miner are you using? Poclbm? It does long polls using HTTP GET requests instead of JSON-RPC, which I didn't know.

EDIT: The latest commit has a fix for this.

Ufasoft bitcoin miner.
hero member
Activity: 516
Merit: 643
Only seems to appear when there is "new work for worker".
New version. Bitcoinqt 0.5.1 on Win7x64.

What miner are you using? Poclbm? It does long polls using HTTP GET requests instead of JSON-RPC, which I didn't know.

EDIT: The latest commit has a fix for this.
Pages:
Jump to: