Author

Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 443. (Read 2591964 times)

legendary
Activity: 1540
Merit: 1001
Quote
You don't have to ask here, the information is already public. Go to p2pool.info and you'll have data for a theoretical 100% efficient p2pool node.

I don't care what theoretical is.  I want to know what people are really getting! Smiley

They are getting what the blocks found by p2pool pay them. These blocks aren't theoretical only a 100% efficient node is.

If your node is averaging around 103% efficiency like mine for example you will get 103% of what p2pool.info reports. That's pure math.

Yes, I understand that.  I still want to see some numbers! Smiley You keep claiming you do better in p2pool than a conventional pool.  Let's see numbers to prove it!  We know theory and reality usually don't meet.

M
full member
Activity: 932
Merit: 100
arcs-chain.com
I'm running a node for testing right now in a 2.6 Ghz Core i7 16GB RAM PCI-e SSD...and Ubuntu Server 12.04

My question is, I'm getting periodic spikes in both the getblocktemploate latency (upwards of 30s for a spike) and DOAs reported up from 4% to almost 30% then it calms down and I get a share.  I have 17 shares 1 orphan 117% Efficiency.  I've got about 1.4TH/s on the node. When this happens the p2pool app pegs CPU and then slowly calms down. I've tuned bitcoind for max block size etc. from the tuning guide as well. I have 6 out and 8 in p2pool connections.

I've also noticed my bitcoind has 98 connections and some of them have huge Recv-Q and Send-Q numbers... I'm on a 50 down 5 up cable modem so the connection shouldn't be the problem.  I'm thinking those "dead" connections to bitcoind aren't doing me any good.

My payouts are going up for each share found, but the Spikes in DOA don't seem right.  Any ideas what could be happening?

I think that with your 5mbit upload the most you can handle is 10-15 bitcoind connections, more if you choose them carefully. Those random connections eats your upload bandwith - because theres allways someone downloading blockchain..
full member
Activity: 216
Merit: 100
Don't let the nam-shub in your operating system.
Igor,

I get the same error and the same spikes in getblocktemplate latency you're getting.  And it seems getting a share isn't related to it as I originally thought.

-Rick
newbie
Activity: 18
Merit: 0
Code:
Share check failed:
2014-01-18 21:03:38.148653 > Traceback (most recent call last):
2014-01-18 21:03:38.148709 >   File "/home/breeba/p2pool/p2pool/p2p.py", line 283, in handle_shares
2014-01-18 21:03:38.148784 >     self.node.handle_shares(result, self)
2014-01-18 21:03:38.148835 >   File "/home/breeba/p2pool/p2pool/node.py", line 48, in handle_shares
2014-01-18 21:03:38.148890 >     self.node.set_best_share()
2014-01-18 21:03:38.148946 >   File "/home/breeba/p2pool/p2pool/node.py", line 295, in set_best_share
2014-01-18 21:03:38.149002 >     best, desired, decorated_heads = self.tracker.think(self.get_height_rel_highest, self.bitcoind_work.value['previous_block'], self.bitcoind_work.value['bits'], self.known_txs_var.value)
2014-01-18 21:03:38.149056 >   File "/home/breeba/p2pool/p2pool/data.py", line 452, in think
2014-01-18 21:03:38.149113 >     if self.attempt_verify(share):
2014-01-18 21:03:38.149163 > --- ---
2014-01-18 21:03:38.149217 >   File "/home/breeba/p2pool/p2pool/data.py", line 431, in attempt_verify
2014-01-18 21:03:38.149271 >     share.check(self)
2014-01-18 21:03:38.149326 >   File "/home/breeba/p2pool/p2pool/data.py", line 320, in check
2014-01-18 21:03:38.149380 >     raise ValueError('share_info invalid')
2014-01-18 21:03:38.149436 > exceptions.ValueError: share_info invalid

I'm getting this error on p2pool and my cpu goes 100%. The error is continuos. I've tried to delete the data folder, then it works nice for few hours and baam the error starts again!

Another doubt: I've set up the pool to collect 2%, why may pout amount is always 0 and I just see payments eery 4 or 5 days?

Any ideas?!

Thank you,
Igor

my node: bitcoin.breeba.com:3330
hero member
Activity: 630
Merit: 501
I'm running a node for testing right now in a 2.6 Ghz Core i7 16GB RAM PCI-e SSD...and Ubuntu Server 12.04

My question is, I'm getting periodic spikes in both the getblocktemploate latency (upwards of 30s for a spike) and DOAs reported up from 4% to almost 30% then it calms down and I get a share.  I have 17 shares 1 orphan 117% Efficiency.  I've got about 1.4TH/s on the node. When this happens the p2pool app pegs CPU and then slowly calms down. I've tuned bitcoind for max block size etc. from the tuning guide as well. I have 6 out and 8 in p2pool connections.

I've also noticed my bitcoind has 98 connections and some of them have huge Recv-Q and Send-Q numbers... I'm on a 50 down 5 up cable modem so the connection shouldn't be the problem.  I'm thinking those "dead" connections to bitcoind aren't doing me any good.

My payouts are going up for each share found, but the Spikes in DOA don't seem right.  Any ideas what could be happening?

How long have you been running for?

I haven't been running for very long, a about 6 hours. But my DOA is 3% at 276GH/s my I have only have 2 shares but efficiency is 129%. I have been hitting 30s a couple of times but it doesn't seem to affect me much. I am using a Xi3 Z3RO Pro computer (4GB Ram, 120GB mSATA, dual core 1.65GHz AMD) running openSUSE not Ubuntu. The only other thing I have differently on my network is my CISCO firewall, managed switch, and P2Pool server have Jumbo Frames enabled. I use jumbo frames on other pieces of hardware in my network, such as my SAN and ESXi server so I figured I would give a try on the Z3RO Pro.

full member
Activity: 216
Merit: 100
Don't let the nam-shub in your operating system.
I'm running a node for testing right now in a 2.6 Ghz Core i7 16GB RAM PCI-e SSD...and Ubuntu Server 12.04

My question is, I'm getting periodic spikes in both the getblocktemploate latency (upwards of 30s for a spike) and DOAs reported up from 4% to almost 30% then it calms down and I get a share.  I have 17 shares 1 orphan 117% Efficiency.  I've got about 1.4TH/s on the node. When this happens the p2pool app pegs CPU and then slowly calms down. I've tuned bitcoind for max block size etc. from the tuning guide as well. I have 6 out and 8 in p2pool connections.

I've also noticed my bitcoind has 98 connections and some of them have huge Recv-Q and Send-Q numbers... I'm on a 50 down 5 up cable modem so the connection shouldn't be the problem.  I'm thinking those "dead" connections to bitcoind aren't doing me any good.

My payouts are going up for each share found, but the Spikes in DOA don't seem right.  Any ideas what could be happening?
hero member
Activity: 896
Merit: 1000
Quote
You don't have to ask here, the information is already public. Go to p2pool.info and you'll have data for a theoretical 100% efficient p2pool node.

I don't care what theoretical is.  I want to know what people are really getting! Smiley

They are getting what the blocks found by p2pool pay them. These blocks aren't theoretical only a 100% efficient node is.

If your node is averaging around 103% efficiency like mine for example you will get 103% of what p2pool.info reports. That's pure math.
hero member
Activity: 630
Merit: 501
is there a users page add-on for p2pool other than the front-end add-on?

Thanks,
hero member
Activity: 630
Merit: 501
I've never seen anyone say good things about QoS.  In fact, I've seen lots of people say turn it off, it hoses things, and I've seen it here in this thread.

I really don't care about what people say but about what is doable. I've configured QoS on my Linux server, it works, end of story. If people use QoS on their Linux server and it doesn't work for them I can provide tips.

That might be the difference.  The average person isn't going to use QoS on a Linux server.  The average person is going to have a cheap router that touts QoS, but the reality is it doesn't work well.

Who doesn't use the bandwidth for anything else?

You misunderstood. QoS allows bandwidth for other traffic but for it to work you must have some headroom, if you need all of your available bandwidth for other traffics you can't use p2pool, if you only need what's above 256-512kbps in both directions then you can fit p2pool and bitcoind in with various levels of efficiency.

I know what QoS is.  I have theoretical 768k up.  If I need p2pool and bitcoin to use 2/3's of that, I'm in trouble.

Quote
You don't have to ask here, the information is already public. Go to p2pool.info and you'll have data for a theoretical 100% efficient p2pool node.

I don't care what theoretical is.  I want to know what people are really getting! Smiley

Quote
Given that you mentioned a cube I suppose you have an ASICMiner cube. Unless they have a different miner implementation than the blades you are probably out of luck on p2pool. IIRC the blade miner could only do getwork and poorly: it didn't work well on several pools including p2pool because of their implementation. If this didn't change you are probably better off letting the beast mine on whatever pools you could find that worked fine with it.

If I was to attempt it I'd run it through a stratum proxy, as I'm doing now.  But as I've implied a few times, there's no way I'm pointing 38gh/s at 1m share difficulty.

M

I would have to say if you have a good switch and aren't using a mickey mouse home one you should be good to go. Home managed switches have very low forwarding rates thus they can't move higher volumes of traffic over them compared to the enterprise class managed switches. So if you a good switch and are using good cabling, I would certainly enable jumbo frames before messing with the QOS. My firewall, Dell PowerConnect 6224 and my openSUSE (p2pool server) have jumbo frames enabled.
legendary
Activity: 1540
Merit: 1001
I've never seen anyone say good things about QoS.  In fact, I've seen lots of people say turn it off, it hoses things, and I've seen it here in this thread.

I really don't care about what people say but about what is doable. I've configured QoS on my Linux server, it works, end of story. If people use QoS on their Linux server and it doesn't work for them I can provide tips.

That might be the difference.  The average person isn't going to use QoS on a Linux server.  The average person is going to have a cheap router that touts QoS, but the reality is it doesn't work well.

Who doesn't use the bandwidth for anything else?

You misunderstood. QoS allows bandwidth for other traffic but for it to work you must have some headroom, if you need all of your available bandwidth for other traffics you can't use p2pool, if you only need what's above 256-512kbps in both directions then you can fit p2pool and bitcoind in with various levels of efficiency.

I know what QoS is.  I have theoretical 768k up.  If I need p2pool and bitcoin to use 2/3's of that, I'm in trouble.

Quote
You don't have to ask here, the information is already public. Go to p2pool.info and you'll have data for a theoretical 100% efficient p2pool node.

I don't care what theoretical is.  I want to know what people are really getting! Smiley

Quote
Given that you mentioned a cube I suppose you have an ASICMiner cube. Unless they have a different miner implementation than the blades you are probably out of luck on p2pool. IIRC the blade miner could only do getwork and poorly: it didn't work well on several pools including p2pool because of their implementation. If this didn't change you are probably better off letting the beast mine on whatever pools you could find that worked fine with it.

If I was to attempt it I'd run it through a stratum proxy, as I'm doing now.  But as I've implied a few times, there's no way I'm pointing 38gh/s at 1m share difficulty.

M
hero member
Activity: 896
Merit: 1000
I've never seen anyone say good things about QoS.  In fact, I've seen lots of people say turn it off, it hoses things, and I've seen it here in this thread.

I really don't care about what people say but about what is doable. I've configured QoS on my Linux server, it works, end of story. If people use QoS on their Linux server and it doesn't work for them I can provide tips.

Who doesn't use the bandwidth for anything else?
See my prior post.  What's your 12 day comparison?  I realize it's probably too short, but I'd like to see.  We can compare again in 18 days.

You don't have to ask here, the information is already public. Go to p2pool.info and you'll have data for a theoretical 100% efficient p2pool node.

Given that you mentioned a cube I suppose you have an ASICMiner cube. Unless they have a different miner implementation than the blades you are probably out of luck on p2pool. IIRC the blade miner could only do getwork and poorly: it didn't work well on several pools including p2pool because of their implementation. If this didn't change you are probably better off letting the beast mine on whatever pools you could find that worked fine with it.
hero member
Activity: 630
Merit: 501
How do I get the p2pool pages for /users/ /local_stats/ etc looking pleasing instead of lines of word wrapped text?

Also I have the latest version installed and /graphs/ doesn't work telling me "no such resource" is this normal?

Thanks


https://github.com/hardcpp/P2PoolExtendedFrontEnd

Sorry when it says put these files in your web-static folder, does it just mean the files or do I need to download the folders too? (just wondering so I am not doing something tiwce)

$ cd p2pool
$ mv web-static web-static.orig
$ unzip P2PoolExtendedFrontEnd.zip
$ mv P2PoolExtendedFrontEnd web-static

from browser hit you 9332 port, you should see improved UI.

-alan

To edit some of the links on the page that's all in the index I suppose?
hero member
Activity: 630
Merit: 501
EDIT: I guess my point is p2pool isn't for everyone.  True supporters of the p2pool concept acknowledge that.  The rest treat p2pool as a religion, as from their perspective, anyone who has a problem must not have their settings right or not understand the system.

On a related subject variance importance is often blown out of proportions and people willingly choose pools that pay between 3 and 5% less monthly because p2pool pays less nearly one day out of two. This behavior is not limited to pool choice: many forfeit long term gains for short term gains in most domains.

I agree, a mining friend of mine told me I should give P2Pool a chance for 30 days or so to see the overall gain over using a fee based pool. So far I have to say at I am happy to be at least hashing, the fee based pool I have been using lately hasn't had any blocks to be hashed so my miners have been spinning their wheels. Sad
legendary
Activity: 1540
Merit: 1001
EDIT: I guess my point is p2pool isn't for everyone.  True supporters of the p2pool concept acknowledge that.  The rest treat p2pool as a religion, as from their perspective, anyone who has a problem must not have their settings right or not understand the system.

Given that p2pool can be configured to use less than 20KB/s and bitcoind tuned to use even less, people with DSL should all be able to use p2pool. The only ways someone can't use p2pool on a link that provides at least 512kbps in both directions I can think of are:
  • not knowing how to configure p2pool and bitcoind for low bandwidth,
  • not knowing how or being able to use QoS to give priority to p2pool and bitcoind,
  • needing the bandwidth for something else,
  • unreliable connectivity

I've never seen anyone say good things about QoS.  In fact, I've seen lots of people say turn it off, it hoses things, and I've seen it here in this thread.

Who doesn't use the bandwidth for anything else?

Quote
On a related subject variance importance is often blown out of proportions and people willingly choose pools that pay between 3 and 5% less monthly because p2pool pays less nearly one day out of two. This behavior is not limited to pool choice: many forfeit long term gains for short term gains in most domains.

See my prior post.  What's your 12 day comparison?  I realize it's probably too short, but I'd like to see.  We can compare again in 18 days.

M
legendary
Activity: 1540
Merit: 1001
Just for comparison... since the common belief is that the high variance and high share difficulty doesn't affect your payout, let's compare.

I have 38GH/s.  It's a cube, I brought it online almost 13 days ago, pointing to BTC Guild.  Since then my total payout has been 0.12764.  According to BTCGuild stats, we're at 99.7% luck for the last 14 days, so I'm pretty close to "average" payout.  I'm not counting the ~0.21 NMC I've earned.

Can someone who has significantly larger hashrate post their comparison?  Ie, what you made in the last 13 days scaled down to 38GH/s?

EDIT: I'd also like to see someone who has an equivalent hashrate or less.

M

hero member
Activity: 896
Merit: 1000
EDIT: I guess my point is p2pool isn't for everyone.  True supporters of the p2pool concept acknowledge that.  The rest treat p2pool as a religion, as from their perspective, anyone who has a problem must not have their settings right or not understand the system.

There are people with a setup that can't work well with p2pool whatever they try (ASICMiner blades, clogged or unreliable Internet connectivity for example) but there are people who don't understand how to tune a p2pool setup to work well on their environment too.

Given that p2pool can be configured to use less than 20KB/s and bitcoind tuned to use even less, people with DSL should all be able to use p2pool. The only ways someone can't use p2pool on a link that provides at least 512kbps in both directions I can think of are:
  • not knowing how to configure p2pool and bitcoind for low bandwidth,
  • not knowing how or being able to use QoS to give priority to p2pool and bitcoind,
  • needing the bandwidth for something else,
  • unreliable connectivity

On a related subject variance importance is often blown out of proportions and people willingly choose pools that pay between 3 and 5% less monthly because p2pool pays less nearly one day out of two. This behavior is not limited to pool choice: many forfeit long term gains for short term gains in most domains.
newbie
Activity: 16
Merit: 0
How do I get the p2pool pages for /users/ /local_stats/ etc looking pleasing instead of lines of word wrapped text?

Also I have the latest version installed and /graphs/ doesn't work telling me "no such resource" is this normal?

Thanks


https://github.com/hardcpp/P2PoolExtendedFrontEnd

Sorry when it says put these files in your web-static folder, does it just mean the files or do I need to download the folders too? (just wondering so I am not doing something tiwce)

$ cd p2pool
$ mv web-static web-static.orig
$ unzip P2PoolExtendedFrontEnd.zip
$ mv P2PoolExtendedFrontEnd web-static

from browser hit you 9332 port, you should see improved UI.

-alan
hero member
Activity: 630
Merit: 501
How do I get the p2pool pages for /users/ /local_stats/ etc looking pleasing instead of lines of word wrapped text?

Also I have the latest version installed and /graphs/ doesn't work telling me "no such resource" is this normal?

Thanks


https://github.com/hardcpp/P2PoolExtendedFrontEnd

Sorry when it says put these files in your web-static folder, does it just mean the files or do I need to download the folders too? (just wondering so I am not doing something tiwce)
legendary
Activity: 1540
Merit: 1001
I pay what I want to forrestv.
P2Pool has a negligible amount of orphans (didn't even find one in the last 200 known p2pool blocks).
I get merged mining with NMC, IXC, DVC (and could add any other merge-able coin I deem worth it).
Variance is only a problem if you don't earn much on p2pool (maybe then it doesn't really matter where you mine). You can actually verify this with merged mined coins: they work OK but as it's solo mining variance is really high and NMC blocks are a rare but nice bonus.
Stales % doesn't matter on p2pool only efficiency and I'm at 100+%.
p2pool pays transaction fees.

And I run it on a DSL line...

YMMV.  My DSL line can't take the bandwidth.  I seem you recall you saying earlier (I think it was you) that p2pool doesn't run well on the average home DSL line.

Remember the 100+% efficiency is compared to other p2pool users, not compared to conventional pools.  I'm sure with a little tweaking I could get below the current p2pool average of 19%.  It's still magnitudes more than I get on a conventional pool.

Quote
I only use centralized pools as backups and really don't see how I would benefit from switching to any of them. Even if I had a whole week of downtime right now on p2pool I still would have earned more than on any theoretical 100% up centralized pool (which actually have downtime when they are DoSed).

BTC Guild had a huge DDOS not that long ago.  It stayed up.

Quote
By the way, look at the current IXC and DVC difficulty (43% and 64% of BTC difficulty). This means that at least 40% of the miners are merge-mining IXC and 60% are merge-mining DVC (I'm pretty sure nobody is mining these coins expecting a profit unless they are using merge-mining).
Do you think these miners are aware they do mine IXC and/or DVC and get the relevant income?

No : they are mined by centralized pools and most keep the income for themselves and AFAIK don't even tell their miners about it. When you use p2pool you can at least fully control what your hash power is used for.

That's an interesting point.

It doesn't mean I can successfully merge mine them though.  Did you see what I posted for IXC, DVC, and NMC difficulty?  It's like playing the lottery trying to solo mine them, which is what you're doing if you merge mine them with p2pool.  I'm better off not merge mining them and save some bandwidth.

You seem to distrust pool operators.  I happen to trust mine, and I am certain I do better on BTC Guild than I would on p2pool.  If I had the bandwidth and more hashpower, I'd use p2pool.  But I don't... I've been effectively squeezed out.  And I'm not alone.

M

EDIT: I guess my point is p2pool isn't for everyone.  True supporters of the p2pool concept acknowledge that.  The rest treat p2pool as a religion, as from their perspective, anyone who has a problem must not have their settings right or not understand the system.
newbie
Activity: 16
Merit: 0
How do I get the p2pool pages for /users/ /local_stats/ etc looking pleasing instead of lines of word wrapped text?

Also I have the latest version installed and /graphs/ doesn't work telling me "no such resource" is this normal?

Thanks


https://github.com/hardcpp/P2PoolExtendedFrontEnd
Jump to: