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

[spiccioli]

I'd like to stress out what

https://bitcointalk.org/index.php?topic=77000.msg1112365#msg1112365

tells about p2pool!

It has the 'L' word in it...

spiccioli

That chart isn't entirely accurate.  Things went sour when the extra 200g/h was added.  Yes, we did have one ridiculously good day when we got like 10 blocks, but most other days were not as good as they should have been when we were almost double the hash rate.

Since that user left, things have returned to normal.

I do think we have a scaling issue.

M

I'm not so sure, a week is not enough to know, p2pool took 8 month to have a 90 days positive pool luck index.

spiccioli

"6 weeks of no bad L" ... except for that period when we had an extra 200g/h.  Seems pretty convincing to me.

M

No, it is "6 weeks of no bad L" including that period.

Maybe those extra 200 GH/s made p2pool bad luck last only a week instead of a month... who can say for sure?

spiccioli

[kjj]

"6 weeks of no bad L" ... except for that period when we had an extra 200g/h.  Seems pretty convincing to me.

You weren't paying attention.  The run of bad luck started before the rate jumped.

While it is still possible that we had a little initial bad luck, then luck returned to normal at the exact moment that we had a scaling problem, it is quite a stretch to call that convincing.

[mdude77]

"6 weeks of no bad L" ... except for that period when we had an extra 200g/h.  Seems pretty convincing to me.

You weren't paying attention.  The run of bad luck started before the rate jumped.

While it is still possible that we had a little initial bad luck, then luck returned to normal at the exact moment that we had a scaling problem, it is quite a stretch to call that convincing.

Apparently the owner of that hash rate was convinced enough.

M

[organofcorti]

I'd like to stress out what

https://bitcointalk.org/index.php?topic=77000.msg1112365#msg1112365

tells about p2pool!

It has the 'L' word in it...

spiccioli

That chart isn't entirely accurate.  Things went sour when the extra 200g/h was added.  Yes, we did have one ridiculously good day when we got like 10 blocks, but most other days were not as good as they should have been when we were almost double the hash rate.

I take comments like this seriously. If you show me and explain exactly which data is inaccurate and I'll fix it immediately.

[mdude77]

That chart isn't entirely accurate.  Things went sour when the extra 200g/h was added.  Yes, we did have one ridiculously good day when we got like 10 blocks, but most other days were not as good as they should have been when we were almost double the hash rate.

I take comments like this seriously. If you show me and explain exactly which data is inaccurate and I'll fix it immediately.

I'm sorry, maybe I'm misunderstanding, and I probably worded it poorly.  The chart is probably fine.  It's your quote about "no bad luck in 6 weeks".  I'm guessing you're looking at the large scheme of things?  We had three instances in the last 14 days where it took > 24 hours for a block.  7 day luck dropped below 90% not that long ago.  What exactly do you mean by no bad luck? 

M

[organofcorti]

That chart isn't entirely accurate.  Things went sour when the extra 200g/h was added.  Yes, we did have one ridiculously good day when we got like 10 blocks, but most other days were not as good as they should have been when we were almost double the hash rate.

I take comments like this seriously. If you show me and explain exactly which data is inaccurate and I'll fix it immediately.

I'm sorry, maybe I'm misunderstanding, and I probably worded it poorly.  The chart is probably fine.  It's your quote about "no bad luck in 6 weeks".  I'm guessing you're looking at the large scheme of things?  We had three instances in the last 14 days where it took > 24 hours for a block.  7 day luck dropped below 90% not that long ago.  What exactly do you mean by no bad luck? 

M

I follow. You're looking at the 7 day rolling mean at p2Pool.info? That's not quite the same as the weekly average chart I produce. Mine is just the week's average luck. It's more useful than the 7 day rolling mean when analysing long term trends. If you average the last 7 weeks data, the average round shares / difficulty is less than 1. This means better than average luck. But the weekly averages don't have the granularity to find small intraweek drops in luck, for which a 7 day rolling mean is more suited.

Even so, for investigating what you're interested in neither is as good a simple correlation analysis. I did that when I wrote the p2Pool Neighbourhood Pool Watch post and didn't find a correlation between hashrate and round length. I can do it again if miners here really think there's a problem. I honestly can't see how increasing the hashrate would increase the probability of more than averge shares to solve a round though. An increase in orphans maybe, and even then there wasn't a significant correlation when last I checked.

 

[mdude77]

I'm sorry, maybe I'm misunderstanding, and I probably worded it poorly.  The chart is probably fine.  It's your quote about "no bad luck in 6 weeks".  I'm guessing you're looking at the large scheme of things?  We had three instances in the last 14 days where it took > 24 hours for a block.  7 day luck dropped below 90% not that long ago.  What exactly do you mean by no bad luck? 

M

I follow. You're looking at the 7 day rolling mean at p2Pool.info? That's not quite the same as the weekly average chart I produce. Mine is just the week's average luck. It's more useful than the 7 day rolling mean when analysing long term trends. If you average the last 7 weeks data, the average round shares / difficulty is less than 1. This means better than average luck. But the weekly averages don't have the granularity to find small intraweek drops in luck, for which a 7 day rolling mean is more suited.

Even so, for investigating what you're interested in neither is as good a simple correlation analysis. I did that when I wrote the p2Pool Neighbourhood Pool Watch post and didn't find a correlation between hashrate and round length. I can do it again if miners here really think there's a problem. I honestly can't see how increasing the hashrate would increase the probability of more than averge shares to solve a round though. An increase in orphans maybe, and even then there wasn't a significant correlation when last I checked.

So, you're saying long term analysis indicates that p2pool is coming out ahead, but no one else is?  I was always under the impression that every pool, if you look at it long enough, will balance out to "in luck".

M

[organofcorti]

I'm sorry, maybe I'm misunderstanding, and I probably worded it poorly.  The chart is probably fine.  It's your quote about "no bad luck in 6 weeks".  I'm guessing you're looking at the large scheme of things?  We had three instances in the last 14 days where it took > 24 hours for a block.  7 day luck dropped below 90% not that long ago.  What exactly do you mean by no bad luck? 

M

I follow. You're looking at the 7 day rolling mean at p2Pool.info? That's not quite the same as the weekly average chart I produce. Mine is just the week's average luck. It's more useful than the 7 day rolling mean when analysing long term trends. If you average the last 7 weeks data, the average round shares / difficulty is less than 1. This means better than average luck. But the weekly averages don't have the granularity to find small intraweek drops in luck, for which a 7 day rolling mean is more suited.

Even so, for investigating what you're interested in neither is as good a simple correlation analysis. I did that when I wrote the p2Pool Neighbourhood Pool Watch post and didn't find a correlation between hashrate and round length. I can do it again if miners here really think there's a problem. I honestly can't see how increasing the hashrate would increase the probability of more than averge shares to solve a round though. An increase in orphans maybe, and even then there wasn't a significant correlation when last I checked.

So, you're saying long term analysis indicates that p2pool is coming out ahead, but no one else is?  I was always under the impression that every pool, if you look at it long enough, will balance out to "in luck".

M

I'm saying that the average (round length)/difficulty < 1 for the last 7 weeks. If you want to try for yourself, take the total shares per round, divide by difficulty and take the mean.

[mdude77]

What's up with block 194900?

Looks like we found it twice, and of now (43 minutes after the last one), still haven't been payed for it?

Double orphan on the same block?

M

[twmz]

What's up with block 194900?

Looks like we found it twice, and of now (43 minutes after the last one), still haven't been payed for it?

Double orphan on the same block?

M

One of them is orphaned and the other is not.  It's a rare glitch on p2pool.info that causes it to not appear orphaned until 2 hours later only when the pool finds two blocks with the same height that are competing.  Later tonight, it will fix itself and one of them will appear orphaned.

[Syke]

I do think we have a scaling issue.

I'm not so sure, a week is not enough to know, p2pool took 8 month to have a 90 days positive pool luck index.

spiccioli

"6 weeks of no bad L" ... except for that period when we had an extra 200g/h.  Seems pretty convincing to me.

M

So when I toss a coin in the air and call out "HEADS", and it lands on heads, that's pretty convincing that I can control a coin toss?

Look, shares start out at 10 seconds on average. When the hashrate goes up, difficulty goes up, and share rate goes to ... 10 seconds on average. "bad luck" is not caused by an increased hashrate.

[mdude77]

What's up with block 194900?

Looks like we found it twice, and of now (43 minutes after the last one), still haven't been payed for it?

Double orphan on the same block?

M

One of them is orphaned and the other is not.  It's a rare glitch on p2pool.info that causes it to not appear orphaned until 2 hours later only when the pool finds two blocks with the same height that are competing.  Later tonight, it will fix itself and one of them will appear orphaned.

We orphaned our own block though?  And I still get payout for that block..

M

[twmz]

What's up with block 194900?

Looks like we found it twice, and of now (43 minutes after the last one), still haven't been payed for it?

Double orphan on the same block?

M

One of them is orphaned and the other is not.  It's a rare glitch on p2pool.info that causes it to not appear orphaned until 2 hours later only when the pool finds two blocks with the same height that are competing.  Later tonight, it will fix itself and one of them will appear orphaned.

We orphaned our own block though?  And I still get payout for that block..

M

yes, we had two blocks in competition for a while.  It looks like both were actually extended by others and it wasn't until the 194902 height that one of them "won" and the other was orhpaned.  You should have gotten a payment for the one that "won" (assuming you had at least 1 share prior to the block being found) and not for the orphaned block.  However, you may have a transaction with 0 confirmations in your transaction log for the orphaned block.  That's normal. It will never get confirmed and so you might as well just ignore it.

[mdude77]

We orphaned our own block though?  And I still get payout for that block..

M

yes, we had two blocks in competition for a while.  It looks like both were actually extended by others and it wasn't until the 194902 height that one of them "won" and the other was orhpaned.  You should have gotten a payment for the one that "won" (assuming you had at least 1 share prior to the block being found) and not for the orphaned block.  However, you may have a transaction with 0 confirmations in your transaction log for the orphaned block.  That's normal. It will never get confirmed and so you might as well just ignore it.

I ran "bitcoind -rescan" and it found the missing transaction.

M

[Naelr]

Is it crazy to see 6.3 gigs upload as a result of P2Pool traffic?  BTC and LTC?

[ChanceCoats123]

Is it crazy to see 6.3 gigs upload as a result of P2Pool traffic?  BTC and LTC?

Over what timeframe? Or am I confused about the question? There is definite up and down traffic because of P2pool (sending and receiving information from other nodes), but until we know the time frame that 6.3gb was uploaded, I wouldn't know. I haven't measured my network usage from P2pool though.

[Naelr]

Is it crazy to see 6.3 gigs upload as a result of P2Pool traffic?  BTC and LTC?

Over what timeframe? Or am I confused about the question? There is definite up and down traffic because of P2pool (sending and receiving information from other nodes), but until we know the time frame that 6.3gb was uploaded, I wouldn't know. I haven't measured my network usage from P2pool though.

oops... I thought I put that in there ... 4 days... is what my box says is the uptime.

[Krak]

Is it crazy to see 6.3 gigs upload as a result of P2Pool traffic?  BTC and LTC?

I do know I'd been running into my data cap (150GB per month) almost every month in the recent months that I was running p2pool. I'm not entirely sure that p2pool was the cause though.

[Naelr]

Is it crazy to see 6.3 gigs upload as a result of P2Pool traffic?  BTC and LTC?

I do know I'd been running into my data cap (150GB per month) almost every month in the recent months that I was running p2pool. I'm not entirely sure that p2pool was the cause though.

Well I noticed on my router 2 gigs a day for the last 2 days ..then noticed I left a torrent seeding.. that might be the cause for my internet being shitty today... but looking at stats on my linux box running p2pool

for the last 4 days...

network history 2.7 GiB down 6.6GiB up ... I have done system up dates as well and I haven't uploaded anything from that computer I am watching it in real time and the outgoing spikes to 250 KiB/s for about 10-15 seconds every minute.   No other services on that box that would send info to the internet ... and no one is connected to my pools....

[Smoovious]

Is it crazy to see 6.3 gigs upload as a result of P2Pool traffic?  BTC and LTC?

I do know I'd been running into my data cap (150GB per month) almost every month in the recent months that I was running p2pool. I'm not entirely sure that p2pool was the cause though.

Maybe check on the # of connections you guys are allowing, both in p2pool, and in your daemons, and cut them down more.

That should drop the cumulative bandwidth used.

-- Smoov

ps: You have a data cap? ... Does the data cap only count on traffic that goes outside of our network? Do you get free in-network traffic?