Page 4 of 7
Re: Downtime this morning
Posted: Thu Jul 20, 2017 1:46 pm
by lxsavard
virtuado wrote:It would be nice to have at least an explanation, there is a lot of discontent suddenly. I think there is lack of proper communication in this matter, no status updates, no news...nothing. One day failure is acceptable, but every day?
To be honest...i liked this format, but in the few time i've been here, to many disconections.
It's true now it's a everyday trouble... I love also the mining pool but now I start to be exhausted of that...
Re: Downtime this morning
Posted: Thu Jul 20, 2017 1:48 pm
by lxsavard
Can you confirm us when the mining pool will run normally because now the system have trouble everyday
Re: Downtime this morning
Posted: Fri Jul 21, 2017 2:49 pm
by jaybizz
Chris Sokolowski wrote:CritterDog wrote:FRISKIE wrote: Do we have data to support the conclusion that NH users are pouring in a high amount of hashrate?
I would like to know this also. What is the source of the problem. Pool has been working fine and now all sudden problems? Is this from big NH orders overloading the system?
The cause is that there are a few X11 users who occasionally connect and then rapidly submit thousands of 0.1 difficulty shares per second. I don't know if it's from Nicehash or somewhere else. We normally process about 15,000 shares per second, but these X11 spikes can bring that over 35,000 shares per second and overload the server.
Previously, the database was the limiting factor since we couldn't insert shares fast enough. But now that we've improved the database, the limitation is the mining server processing all these shares. We need to make a quick fix to reduce the number of shares, and I think that setting a cap on per-worker share submission rate has the least effect on users. When Steve reworks the mining server to handle more shares, we can relax the limitation so that it affect only the most egregious violations.
I'm assuming this is going to be worked on Saturday then? Because there's been a ton of these issues for the past several days. A quick fix, IMO, would have been to disable X11 mining so the bulk of your miners wouldn't be put off by this problem in the meantime.
Re: Downtime this morning
Posted: Fri Jul 21, 2017 2:55 pm
by Steve Sokolowski
lxsavard wrote:Can you confirm us when the mining pool will run normally because now the system have trouble everyday
Unfortunately not. Every time we increase capacity, more miners arrive.
Once this issue is fixed, it's likely there will be another one. We hope to get ahead of this someday, and we were for about a week, but there's a limit to how fast we can develop. It's possible that we may reach a point where we have satisfied all possible load, but I don't think that will happen for years.
Re: Downtime this morning
Posted: Fri Jul 21, 2017 3:36 pm
by FRISKIE
Close registration for a while >> get ahead of the curve >> then re-open!
If the issue returns >> repeat
Simple, right?
FRISKIE
Re: Downtime this morning
Posted: Fri Jul 21, 2017 5:24 pm
by megaquake
this applies to scrypt but x11 is similar
I think the pool needs to enforce difficulty rules, I would suggest that no one should be submitting more than 30 shares per minute, NH users have a min difficulty of 8192 and I am sure that they set that as their static difficulty regardless of the amount of hash that they rent, NH has a 100mh min for orders and 8192 is too low, I would suggest a difficulty setting of 16384 for every 100 mh, if you look at the 100 mh orders at NH they are running near 200 mh all the time which is an attack on the stratum server if 8192 static is set, I know that regular miner users do the same and those rules should be applied to those also, I would contact those users and advise them properly, our profit and reliability would increase substantially if the proper difficulty rules were implemented.
Re: Downtime this morning
Posted: Fri Jul 21, 2017 6:47 pm
by FRISKIE
@ megaquake - I use NH myself and agree that your recommendation sounds sensible.
Re: Downtime this morning
Posted: Sat Jul 22, 2017 2:11 am
by megaquake
Pool online but saying share not accepted no response from pool but my miners are still is connected wasting electricity, ill return when you figure out how to remove the abusive low diff users
Re: Downtime this morning
Posted: Sat Jul 22, 2017 12:35 pm
by CritterDog
Steve Sokolowski wrote:lxsavard wrote:Can you confirm us when the mining pool will run normally because now the system have trouble everyday
Unfortunately not. Every time we increase capacity, more miners arrive.
Once this issue is fixed, it's likely there will be another one. We hope to get ahead of this someday, and we were for about a week, but there's a limit to how fast we can develop. It's possible that we may reach a point where we have satisfied all possible load, but I don't think that will happen for years.
It seems to me it would be a good idea to close registration to new miners only until you get a handle on this.. Then you could have limited new miners join as you get system stable.. Nobody wants to be having problems for years..
Re: Downtime this morning
Posted: Sat Jul 22, 2017 2:09 pm
by Steve Sokolowski
CritterDog wrote:Steve Sokolowski wrote:lxsavard wrote:Can you confirm us when the mining pool will run normally because now the system have trouble everyday
Unfortunately not. Every time we increase capacity, more miners arrive.
Once this issue is fixed, it's likely there will be another one. We hope to get ahead of this someday, and we were for about a week, but there's a limit to how fast we can develop. It's possible that we may reach a point where we have satisfied all possible load, but I don't think that will happen for years.
It seems to me it would be a good idea to close registration to new miners only until you get a handle on this.. Then you could have limited new miners join as you get system stable.. Nobody wants to be having problems for years..
All that will do is to cause accounts to gain value.
The only real solution is to raise fees until enough people leave, and then lower them as capacity increases back to this level. But we don't want to do that just yet because we can make significant improvements pretty quickly, I think.