Author

Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 480. (Read 2591928 times)

hero member
Activity: 686
Merit: 500
WANTED: Active dev to fix & re-write p2pool in C
I'm getting better latency, though I am only merge mining NMC.  Machine is an i5-3570S (the low-power 3.1/3.8GHz), 16GB DDR3-1600, and 2x250GB Samsung 840 SSD.  Running on Windows 8.1.  Machine also runs BFGMiner for all my mining hardware, is my home NAS, downloader machine, VM host...

2 x 250GB Samsung 840 SSD!!  mucho denero amigo...... Wink

I'm using a crappy old Intel 320 series 40GB SSD for the OS & a 120GB HyperX SSD for the wallet data, but my setup is solely for headless mining.

Edit: I'm actually amazed at how good these Sempron 145's are when they are unlocked - with such little power.
hero member
Activity: 1246
Merit: 501
I'm getting better latency, though I am only merge mining NMC.  Machine is an i5-3570S (the low-power 3.1/3.8GHz), 16GB DDR3-1600, and 2x250GB Samsung 840 SSD.  Running on Windows 8.1.  Machine also runs BFGMiner for all my mining hardware, is my home NAS, downloader machine, VM host...
hero member
Activity: 686
Merit: 500
WANTED: Active dev to fix & re-write p2pool in C
Hi guys, I did a little upgrade to my mining rig. I am mining on Linux Xubuntu 12.04 LTS 64bit.
Components:
Code:
AMD A10-5700 APU
FM2A75 Pro4 Motherboard
RAM upgrade from:
Code:
4GB RAM DDR3 1333MHz (PC3-10600) dual channel (2x 2GB) by Samsung
Upgraded to:
Code:
16GB RAM DDR3 2400MHz (PC3-19200) dual channel (2x8GB) by Patriot
Result:

 Grin
All components on stock speeds, no O/C, bitcoind configured to mine 1MB blocks (lots of transactions and Tx Fees).


I dunno lenny dude, it looks and sounds like a mighty fine setup alright. But when I think about it, it's a whole lot of power & expense for little to no performance improvement. Don't get me wrong, it's great - but I would have expected better performance looking at your latency graph (although I do agree with zvs - there's far too much emphasis on latency for some reason, it's not as important as most people think, I think).
Looking at your setup & comparing it to my cheapo budget thing, you've got twice as many cores using 50% more power & twice as much RAM running at 50% higher frequency - but the end result doesn't seem to be much different to my graph:



I'm also merge mining 6 coins - but I think you are too, am I right? Is the 1mb block size the only bitcoind adjustment you've made?

Peace.
zvs
legendary
Activity: 1680
Merit: 1000
https://web.archive.org/web/*/nogleg.com
Just some results from a bunch of earlier posts that I haven't cared much about lately since I don't mine bitcoins anymore

Code:
2013-10-01 03:10:46.189979 Skipping from block 22dc8069705acbdb1702858c4c1af48d2c91a6b47b619422a to block e6f4895ae2f13a64f139e0a3791245654daa9cf69ccfb391c!
2013-10-01 03:10:46.202092 New work for worker! Difficulty: 6.086919 Share difficulty: 85295.846524 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:46.207936 New work for worker! Difficulty: 6.086919 Share difficulty: 85295.846524 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:46.214615 New work for worker! Difficulty: 6.086919 Share difficulty: 85295.846524 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:46.220903 New work for worker! Difficulty: 6.086919 Share difficulty: 85295.846524 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:46.289194 Not punishing share for 'Block-stale detected! height(22dc8069705acbdb1702858c4c1af48d2c91a6b47b619422a) < height(e6f4895ae2f13a64f139e0a3791245654daa9cf69ccfb391c) or 191cdc20 != 191cdc20'! Not jumping from 1e08794f to 460c75f3!
2013-10-01 03:10:48.548898 P2Pool: 17355 shares in chain (17359 verified/17359 total) Peers: 46 (5 incoming)
2013-10-01 03:10:48.548973  Local: 29628MH/s in last 10.0 minutes Local dead on arrival: ~12.1% (9-16%) Expected time to share: 3.4 hours
2013-10-01 03:10:48.549013  Shares: 1 (0 orphan, 0 dead) Stale rate: ~0.0% (0-80%) Efficiency: ~114.2% (23-115%) Current payout: 0.0000 BTC
2013-10-01 03:10:48.549066  Pool: 16743GH/s Stale rate: 12.4% Expected time to block: 10.6 hours
2013-10-01 03:10:48.777949 GOT SHARE! xxxxxxxxxxxxxxx 4a7e16bf prev 1e08794f age 11.30s DEAD ON ARRIVAL
2013-10-01 03:10:48.843855 New work for worker! Difficulty: 6.193214 Share difficulty: 85453.676084 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:48.849668 New work for worker! Difficulty: 6.193214 Share difficulty: 85453.676084 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:48.856205 New work for worker! Difficulty: 6.193214 Share difficulty: 85453.676084 Total block value: 25.000000 BTC including 0 transactions
2013-10-01 03:10:48.861889 New work for worker! Difficulty: 6.193214 Share difficulty: 85453.676084 Total block value: 25.000000 BTC including 0 transactions

(5 minutes later)

2013-10-01 03:15:09.272347 P2Pool: 17365 shares in chain (17369 verified/17369 total) Peers: 46 (5 incoming)
2013-10-01 03:15:09.272418  Local: 30838MH/s in last 10.0 minutes Local dead on arrival: ~12.9% (10-16%) Expected time to share: 3.3 hours
2013-10-01 03:15:09.272446  Shares: 2 (0 orphan, 0 dead) Stale rate: ~0.0% (0-66%) Efficiency: ~116.7% (39-117%) Current payout: 0.0000 BTC
2013-10-01 03:15:09.272477  Pool: 16872GH/s Stale rate: 14.3% Expected time to block: 10.5 hours

'punishing share for block stale detected' == owning yourself for having good hardware/low latency connection.  dont do it
legendary
Activity: 1036
Merit: 1000
DARKNETMARKETS.COM
Hi guys, I did a little upgrade to my mining rig. I am mining on Linux Xubuntu 12.04 LTS 64bit.
Components:
Code:
AMD A10-5700 APU
FM2A75 Pro4 Motherboard
RAM upgrade from:
Code:
4GB RAM DDR3 1333MHz (PC3-10600) dual channel (2x 2GB) by Samsung
Upgraded to:
Code:
16GB RAM DDR3 2400MHz (PC3-19200) dual channel (2x8GB) by Patriot
Result:

 Grin
All components on stock speeds, no O/C, bitcoind configured to mine 1MB blocks (lots of transactions and Tx Fees).
zvs
legendary
Activity: 1680
Merit: 1000
https://web.archive.org/web/*/nogleg.com
Is there any sort of "front end" that you can connect to by name, that finds the fastest/closest p2pool server to you?

The p2pool server daemon itself sort-of does this already, since it connects to multiple other nodes.  Having a node in your own network is the lowest latency solution for the vast majority of people, and even if a front end existed that would find the lowest latency public p2pool node for you, it would take about as much work to set up and maintain as the actual p2pool daemon takes.

I could see a use for this, though, if you couldn't guarantee the availability of your p2pool node even within your own network and didn't want to statically assign a list of backup nodes into your miner.  Still, it's one more point of failure, unnecessary complexity, and you'd still better have a static list in case your auto-connect front end fails.

well, first off, to OP:  fastest =/ closest

well-connected european nodes will have less orphans than nodes in western hemisphere

when I was running a local node, I had an i7-960, 24GB RAM, had the whole bitcoin blockchain stored in RAM, getblocktemplate latency <1ms, but it still performed worse than either of my european servers because of orphan rate

probably not so bad now that shares are farther apart
legendary
Activity: 3430
Merit: 3080
The common garden variety bitcoin miner is a fickle creature; one minute they turn up crying about DOS, next minute they're all gone again, as if DOS has suddenly become impossible. Anyone would think you can ROI your equipment on a Big-Hash-Pool during the 24 hours this mini panic took place.

I'm really looking forward to the return of the days when mining was inhabited by more people that are motivated by Bitcoin's success than by people that are only interested in their own success. And the sad thing is they're either (or both) too lazy or conservative to get that their outcome at p2pool could easily outperform Big-Hash-Pool, but maybe that will come into focus a little more once the transaction rate further saturates the current upper capacity.
sr. member
Activity: 448
Merit: 250
Problem is p2pool is virtually unusable at this level with little guys like me.  12gh/s and estimated time to share is 18.6hours. Sad

Yeah, I know what you mean. That's why I've configured my node as a low-rate-friendly node to reduce the share variance for each individual miner. Check the link in my sig for details.
legendary
Activity: 1379
Merit: 1003
nec sine labore
Problem is p2pool is virtually unusable at this level with little guys like me.  12gh/s and estimated time to share is 18.6hours. Sad

M

There is good news:  The hashrate of the pool is dropping again.  You'll get more shares more often.

However, there seems to be some bad news:  The hashrate of the pool is dropping again.  There won't be as many blocks found as often.

One year ago it was supposed that p2pool had to be split at around 1 TH/s , today it was at 60!

spiccioli

ps. btw, you can always move to HHTT which needs hashers and where you can set your difficulty to a much lower level.
newbie
Activity: 21
Merit: 0
Problem is p2pool is virtually unusable at this level with little guys like me.  12gh/s and estimated time to share is 18.6hours. Sad

M

There is good news:  The hashrate of the pool is dropping again.  You'll get more shares more often.

However, there seems to be some bad news:  The hashrate of the pool is dropping again.  There won't be as many blocks found as often.
newbie
Activity: 21
Merit: 0
Is there any sort of "front end" that you can connect to by name, that finds the fastest/closest p2pool server to you?

The p2pool server daemon itself sort-of does this already, since it connects to multiple other nodes.  Having a node in your own network is the lowest latency solution for the vast majority of people, and even if a front end existed that would find the lowest latency public p2pool node for you, it would take about as much work to set up and maintain as the actual p2pool daemon takes.

I could see a use for this, though, if you couldn't guarantee the availability of your p2pool node even within your own network and didn't want to statically assign a list of backup nodes into your miner.  Still, it's one more point of failure, unnecessary complexity, and you'd still better have a static list in case your auto-connect front end fails.
legendary
Activity: 1540
Merit: 1001
Problem is p2pool is virtually unusable at this level with little guys like me.  12gh/s and estimated time to share is 18.6hours. Sad

M
legendary
Activity: 1379
Merit: 1003
nec sine labore
Is there any sort of "front end" that you can connect to by name, that finds the fastest/closest p2pool server to you?

dlasher, you can start from here http://p2pool-nodes.info/

sorry, I meant from here https://bitcointalksearch.org/topic/updated-current-p2pool-server-list-264533  Smiley


spiccioli
sr. member
Activity: 467
Merit: 250
Quote

thats the benefit from a decentralized Pool like p2poll :-) nobody can hit it.



That's not accurate, it's just more difficult and would cost more time and resources to really DoS the whole P2Pool network. There are many more servers to DoS (DoS and DDoS tools don't know how to iterate through all P2Pool servers this would have to be done separately and distributing the attack on all nodes would have to be done manually). Many servers have weaker Internet connectivity and would be easy to DoS but them being DoSed wouldn't disrupt the rest of the network.

In case of an attack trying to disrupt P2Pool, it would probably be only effective in disrupting a handful of nodes and people with backups would probably not even notice it.

well said.. I'm hoping more and more people put up p2pool nodes, which helps with the survivability.

Is there any sort of "front end" that you can connect to by name, that finds the fastest/closest p2pool server to you?

zvs
legendary
Activity: 1680
Merit: 1000
https://web.archive.org/web/*/nogleg.com
We need a bann-funkction for Clients on public-nodes:

On my node there is a unknown worker wich produces >30% of DOA and massivly
2013-09-30 12:41:24.961081 Worker xxx  submitted share with hash > target:
2013-09-30 12:41:24.961407     Hash:   a9b3716dfc6ad5c4e8145dc2b5529e497e8f45e85cd616d3c6119b00
2013-09-30 12:41:24.961601     Target:  fd1f6b8accdf2000000000000000000000000000000000000000000

what can I /we do again so Workers?
can Dev implement some Score-Function to bann such nodes?

Greets

Subo
well, you could just firewall it yourself

i get a lot of that too, but it doesn't matter to me... i guess it makes your stats look worse, but i've always had fee set to 0% on bitcoin pool anyway
hero member
Activity: 1246
Merit: 501
sr. member
Activity: 344
Merit: 250
Flixxo - Watch, Share, Earn!
BTCGuild is dead due to DDoS, so I'm guessing a lot of the high-power folks have a P2Pool set up as backup.

Edit: It seems slush's, 50BTC and a few other big pools are being nerfed by the big DDoS bat today.

thats the benefit from a decentralized Pool like p2poll :-) nobody can hit it.

That's not accurate, it's just more difficult and would cost more time and resources to really DoS the whole P2Pool network. There are many more servers to DoS (DoS and DDoS tools don't know how to iterate through all P2Pool servers this would have to be done separately and distributing the attack on all nodes would have to be done manually). Many servers have weaker Internet connectivity and would be easy to DoS but them being DoSed wouldn't disrupt the rest of the network.

In case of an attack trying to disrupt P2Pool, it would probably be only effective in disrupting a handful of nodes and people with backups would probably not even notice it.
did You know how many p2pool-nodes exist?
hero member
Activity: 896
Merit: 1000
BTCGuild is dead due to DDoS, so I'm guessing a lot of the high-power folks have a P2Pool set up as backup.

Edit: It seems slush's, 50BTC and a few other big pools are being nerfed by the big DDoS bat today.

thats the benefit from a decentralized Pool like p2poll :-) nobody can hit it.

That's not accurate, it's just more difficult and would cost more time and resources to really DoS the whole P2Pool network. There are many more servers to DoS (DoS and DDoS tools don't know how to iterate through all P2Pool servers this would have to be done separately and distributing the attack on all nodes would have to be done manually). Many servers have weaker Internet connectivity and would be easy to DoS but them being DoSed wouldn't disrupt the rest of the network.

In case of an attack trying to disrupt P2Pool, it would probably be only effective in disrupting a handful of nodes and people with backups would probably not even notice it.
sr. member
Activity: 344
Merit: 250
Flixxo - Watch, Share, Earn!
BTCGuild is dead due to DDoS, so I'm guessing a lot of the high-power folks have a P2Pool set up as backup.

Edit: It seems slush's, 50BTC and a few other big pools are being nerfed by the big DDoS bat today.

thats the benefit from a decentralized Pool like p2poll :-) nobody can hit it.
hero member
Activity: 1246
Merit: 501
BTCGuild is dead due to DDoS, so I'm guessing a lot of the high-power folks have a P2Pool set up as backup.

Edit: It seems slush's, 50BTC and a few other big pools are being nerfed by the big DDoS bat today.
Jump to: