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

[Prelude]

What are ideal share difficulty and pseudo difficulty settings for an S4?

I don't own an S4, nor am I likely to, but I would hazard a guess at the standard 2048 that they work with? Maybe someone who's running one here can chime in?

Peace  

Yeah, I figured 2048 would be ideal for the pseudo difficulty but I'm not sure what to set the other difficulty at? Sorry if this doesn't make sense, I just set up a p2pool node on Linux for the first time and I don't have much experience with p2pool in general.

It seems to be working very well without specifying either, though. 2 units are showing a combined 4.2TH/s and about 3% DOA. I'm just wondering if I can tweak anything else to make things as efficient as possible.

[MissouriMiner]

What are ideal share difficulty and pseudo difficulty settings for an S4?

I don't own an S4, nor am I likely to, but I would hazard a guess at the standard 2048 that they work with? Maybe someone who's running one here can chime in?

Peace  

Yeah, I figured 2048 would be ideal for the pseudo difficulty but I'm not sure what to set the other difficulty at? Sorry if this doesn't make sense, I just set up a p2pool node on Linux for the first time and I don't have much experience with p2pool in general.

It seems to be working very well without specifying either, though. 2 units are showing a combined 4.2TH/s and about 3% DOA. I'm just wondering if I can tweak anything else to make things as efficient as possible.

2048 works well for me.  Your units are doing well.  I also have 2 and only average 3.9 TH/s combined based on cgminer/WebUI output.  I'm not sure what the DOA is for those 2 alone because I have a lot of other miners (mine, with same BTC address) on the pool.  I think I was seeing around 2% DOA when I tested them for a short time by themselves.   If you want to check out my stats, the address is http://bitcoin.missouriminer.com:9332.   I have 2x S4's, 5x S3's, and 2x S2's.  Totaling 8TH/s with 5.9% DOA.  I make about .06 per block.  ( P2Pool found 6 blocks in 1 day a few days ago! )  The S2's are driving most of the DOA, but I still get around 950 GH/s with them so I think they are worth running on the pool.  I have a dedicated business cable network connection for mining, 100Mb down, 4Mb up.  I have 2 separate cable feeds running to my house.  1 business and 1 residential.   Good luck.

[Polyatomic]

2048 works well for me.  Your units are doing well.  I also have 2 and only average 3.9 TH/s combined based on cgminer/WebUI output.  I'm not sure what the DOA is for those 2 alone because I have a lot of other miners (mine, with same BTC address) on the pool.  I think I was seeing around 2% DOA when I tested them for a short time by themselves.   If you want to check out my stats, the address is http://bitcoin.missouriminer.com:9332.   I have 2x S4's, 5x S3's, and 2x S2's.  Totaling 8TH/s with 5.9% DOA.  I make about .06 per block.  ( P2Pool found 6 blocks in 1 day a few days ago! )  The S2's are driving most of the DOA, but I still get around 950 GH/s with them so I think they are worth running on the pool.  I have a dedicated business cable network connection for mining, 100Mb down, 4Mb up.  I have 2 separate cable feeds running to my house.  1 business and 1 residential.   Good luck.

Fully blown sickness man ...

[newbuntu]

@ Ki773r: PatMan is correct. I done some testing of these settings some time ago & posted the results:

I've ordered from batch 1, batch 4 and batch 5.  Even in batch 5 the issues remain, although I did get "lucky" as one of my batch 5 units actually hashes at 504GH/s stable.

I also have one of those lucky ones at 505Gh/s from B1 - it's the most stable one out the lot!!

Following up from this, I've been experimenting a little more with my settings & here's a screen of the results:



(click for larger)

This is 4 x S3's running at various clock speeds. Looking at the graph, I was running --queue 1 up until ~3am (yup, I'm a night time fiddler ) before changing the setting to --queue 0 & letting them run for the same amount of time. It can clearly be seen that after changing the setting to --queue 0, the DOA rate dropped & smoothed out - this was also confirmed by my nodes info page. Average hash rate was slightly higher as a result, so I'll be keeping all my S3's running with the --queue 0 setting from now on. I'm not saying that this will work for everyone, but it's definitely good for my setup & worth giving a try if you're experiencing a higher than expected DOA rate.
The dip in hash rate at the end of the graph was due to a reboot after updating Xubuntu.

Smoke'em if ya got'em  

Edit: It's also worth mentioning that my reject rate was at ~4% with --queue 1 - and ~2% with the 0 setting. This is running on a local node.

You are quite knowledgeable of p2pool, but to suggest that someone "doesn't know anything about what he is doing" when they clearly do, belittles you & your "legendary" status. Seems you owe him an apology  

Since setting --queue 0 my DOA has dropped to 2.6% and my stale rate is only 1 over the last 16 shares, this is much better than I was getting before making the change.
If there are any other helpful tweaks I'd be happy to consider them for my miners.

Also, and to re-ask, anyone know precisely what the "bad peer ban" refers to? - is it a malicious activity or just a miner having a blonde moment?
Cheers.

[PatMan]

It's just a bad (blonde  ) node, nothing malicious. It shouldn't happen too often, but if you're getting a flood of them from the same IP you can block it with IPtables in Ubuntu 

[IYFTech]

@ Ki773r: PatMan is correct. I done some testing of these settings some time ago & posted the results:

I've ordered from batch 1, batch 4 and batch 5.  Even in batch 5 the issues remain, although I did get "lucky" as one of my batch 5 units actually hashes at 504GH/s stable.

I also have one of those lucky ones at 505Gh/s from B1 - it's the most stable one out the lot!!

Following up from this, I've been experimenting a little more with my settings & here's a screen of the results:



(click for larger)

This is 4 x S3's running at various clock speeds. Looking at the graph, I was running --queue 1 up until ~3am (yup, I'm a night time fiddler ) before changing the setting to --queue 0 & letting them run for the same amount of time. It can clearly be seen that after changing the setting to --queue 0, the DOA rate dropped & smoothed out - this was also confirmed by my nodes info page. Average hash rate was slightly higher as a result, so I'll be keeping all my S3's running with the --queue 0 setting from now on. I'm not saying that this will work for everyone, but it's definitely good for my setup & worth giving a try if you're experiencing a higher than expected DOA rate.
The dip in hash rate at the end of the graph was due to a reboot after updating Xubuntu.

Smoke'em if ya got'em  

Edit: It's also worth mentioning that my reject rate was at ~4% with --queue 1 - and ~2% with the 0 setting. This is running on a local node.

You are quite knowledgeable of p2pool, but to suggest that someone "doesn't know anything about what he is doing" when they clearly do, belittles you & your "legendary" status. Seems you owe him an apology  

Since setting --queue 0 my DOA has dropped to 2.6% and my stale rate is only 1 over the last 16 shares, this is much better than I was getting before making the change.

Cheers.

Glad it worked for you! 

[K1773R]

If you have polled upstream and the recommended value for queue is 1, why use 0.

its a fix for a problem years ago. everyone who still uses/recommends it dosnt know anything about what hes doing

i revoke my statement above. seems there is still crappy HW out that needs this setting.

[PatMan]

If you have polled upstream and the recommended value for queue is 1, why use 0.

its a fix for a problem years ago. everyone who still uses/recommends it dosnt know anything about what hes doing

i revoke my statement above. seems there is still crappy HW out that needs this setting.

Was that an apology? 

If it was, I accept it 

[newbuntu]

If you have polled upstream and the recommended value for queue is 1, why use 0.

its a fix for a problem years ago. everyone who still uses/recommends it dosnt know anything about what hes doing

i revoke my statement above. seems there is still crappy HW out that needs this setting.

I'm using S3s. I may experiment with other queue settings to see if there's a noticeable change / difference.

[K1773R]

If you have polled upstream and the recommended value for queue is 1, why use 0.

its a fix for a problem years ago. everyone who still uses/recommends it dosnt know anything about what hes doing

i revoke my statement above. seems there is still crappy HW out that needs this setting.

I'm using S3s. I may experiment with other queue settings to see if there's a noticeable change / difference.

mainly devices with really poor CPU power struggle with the queue (under 100% load), thats why disabling it decreases stales.

[newbuntu]

If you have polled upstream and the recommended value for queue is 1, why use 0.

its a fix for a problem years ago. everyone who still uses/recommends it dosnt know anything about what hes doing

i revoke my statement above. seems there is still crappy HW out that needs this setting.

I'm using S3s. I may experiment with other queue settings to see if there's a noticeable change / difference.

mainly devices with really poor CPU power struggle with the queue (under 100% load), thats why disabling it decreases stales.

That makes a lot of sense, thank you for the explanation.
Cheers.

[mahrens917]

Every once in a while I see a line that says "bad peer banned ... "
can someone please elaborate on what this actually is?
Is this actually someone attempting something nefarious?
Is this a permanent ban for this peer?
Is there a ban list I can manually edit?
Cheers.

EDIT:
I do have firewalls enabled for each device and at each level of the network.

Looking at the code, I believe the list of bad peers is generated in memory so there is no file store of the bad peers.  A peer is added to the list when they submit invalid shares to your node.  Looking at my logs I get a few of these and since they are not from the same IP I do not see them as "nefarious".  Perhaps they could be due to intent or by mistake but since that IP is banned immediately I don't see it as pertinent.

[newbuntu]

Every once in a while I see a line that says "bad peer banned ... "
can someone please elaborate on what this actually is?
Is this actually someone attempting something nefarious?
Is this a permanent ban for this peer?
Is there a ban list I can manually edit?
Cheers.

EDIT:
I do have firewalls enabled for each device and at each level of the network.

Looking at the code, I believe the list of bad peers is generated in memory so there is no file store of the bad peers.  A peer is added to the list when they submit invalid shares to your node.  Looking at my logs I get a few of these and since they are not from the same IP I do not see them as "nefarious".  Perhaps they could be due to intent or by mistake but since that IP is banned immediately I don't see it as pertinent.

Thanks for the input.
It's painful, but i'm sitting watching the terminal window non-stop, any address which is flagged as bad peer banned then I add to my list and block the address and address block in my modem, ufw and router.
I also see other errors crop up and a peer drop and i copy the address into a text file, if i get the same peer more than once then I block that IP as well.
Overkill? perhaps. But I'd rather be safe than sorry.

[PatMan]

You got the bug bad.........

[newbuntu]

You got the bug bad.........

Yep, just finished putting together a dual hex-core xeon server so I can free up my desktop and have something new to sit and stare at - can't wait.
Speaking of that - I wonder how many users it can support?
Is 12GB mem enough for a multi merged server like this?
Already learning lots, having loads of fun.

[PatMan]

12GB should be enough, I'm using 16GB merge mining 8 coins & it's a bit overkill tbh. Mining is also how I learned to use Xubuntu - it's an excellent & fun way to do it eh? 

[newbuntu]

12GB should be enough, I'm using 16GB merge mining 8 coins & it's a bit overkill tbh. Mining is also how I learned to use Xubuntu - it's an excellent & fun way to do it eh? 

It's a great learning curve.
I put Ubuntu Server on it, but I'm struggling with the strict text interface, so I broke down and put ubuntu-desktop on for now - until I am more confident.
I'm just doing install after install of everything from the raid0 setup, bare minimum server install, updates and everything needed for the p2pool server and merge coins.
Get used to it, get comfortable, break it, fix it, break it, fix it, rinse and repeat. I'll be ready to go live with it, perhaps, in a day or so.
Any and all advice welcome.
Cheers.

[mdude77]

12GB should be enough, I'm using 16GB merge mining 8 coins & it's a bit overkill tbh. Mining is also how I learned to use Xubuntu - it's an excellent & fun way to do it eh? 

It's a great learning curve.
I put Ubuntu Server on it, but I'm struggling with the strict text interface, so I broke down and put ubuntu-desktop on for now - until I am more confident.
I'm just doing install after install of everything from the raid0 setup, bare minimum server install, updates and everything needed for the p2pool server and merge coins.
Get used to it, get comfortable, break it, fix it, break it, fix it, rinse and repeat. I'll be ready to go live with it, perhaps, in a day or so.
Any and all advice welcome.
Cheers.

Check out webmin.  Makes managing a text based linux server _so_ much easier.

M

[Prelude]

I'm having a really strange problem with p2pool and I can't figure it out. Something related to it is killing my ethernet connection, seemingly when I refresh the statistics page randomly.

I asked for help on ubuntu forums, since I think it was a ubuntu problem but now I strongly suspect the issue lies with p2pool.

Here is what I posted on ubuntu forums with all the details: http://ubuntuforums.org/showthread.php?t=2249933

Can a linux/p2pool guru give me a hand? I'm pulling my hair out over here and I've tried everything I can think of / find on google. I'm a linux noob.

[newbuntu]

I'm having a really strange problem with p2pool and I can't figure it out. Something related to it is killing my ethernet connection, seemingly when I refresh the statistics page randomly.

I asked for help on ubuntu forums, since I think it was a ubuntu problem but now I strongly suspect the issue lies with p2pool.

Here is what I posted on ubuntu forums with all the details: http://ubuntuforums.org/showthread.php?t=2249933

Can a linux/p2pool guru give me a hand? I'm pulling my hair out over here and I've tried everything I can think of / find on google. I'm a linux noob.

I was having the exact same thing happen to me a few months ago, on two different machines, about 2 weeks apart.

The first time it happened was when I started running with two nics to two different neworks, everytime I restarted the computer both connections start, and they seem to fight over who's in control, (even though they're going out to different networks - probably my lack of knowledge and not setting things up correctly), but, if I turned off one and then restarted the other it would be fine and the connection would no longer drop out. For this machine I never solved it, but the band-aid solution of not using the 2nd connection was adequate.

The second instance was a stand-alone nic in a basic 14.04 desktop environment, the connection kept dropping out, I did some updates and restarted the machine and it stopped happening. I don't know what the actual cure was, but double checking that you have all updates and restarting may provide relief.

Unfortunately, this doesn't give you any hard data. But, you're not alone, I did a quick google and a lot of people have had similar happenings, one was solved by uninstalling amule, one was solved by removing the nic and installing a new compatible one.

Cheers.