Status as of Wednesday, December 13, 2017

[Steve Sokolowski]

Good morning! We have a lot of great news to report today.

• We made significant advances yesterday, all of which were released in two server restarts yesterday. The first restart included a solution to the memory leak that I spent 30 hours tracking down. As a result, the only server restarts now will be those where we are issuing a release to fix a problem.
• The second release resolved an issue with "idle miners." There are some mining rigs that, for an unknown reason, connect to the pool and then never submit shares. There are others that suddenly stop submitting shares but remain connected. To save resources, these miners are marked as "idle" and assigned solely to litecoins until they actually start submitting shares. A miner normally isn't intended to be marked as "idle" until several minutes elapses after connection, but a bug caused miners to be marked idle starting immediately at connection time. Since "idle" miners are considered to not be spending electricity, electricity calculations would be incorrect for customers who connected and disconnected frequently. Since there may still be some bugs that are disconnecting customers, those most affected by those bugs would see very low electricity readings. Incorrectly identified idle miners also affected profitability because they could not be assigned to the most profitable coin.
• The second release also reverted a "fix" I made earlier to the share inserters that incorrectly made the electricity calculations even worse than the errors caused by the "idle miners" issue. Thus, electricity usage calculations, something that we were having a lot of trouble tracking down for a while, appear to finally be corrected and are now immediately available for use.
• We're proud to announce that Chris completed all of the tickets in the old support system. There are two more involving Coinbase payouts, and then the old system will be taken offline. Chris is hoping to be caught up with support tickets by Friday, at which point he'll be able to move onto new exchanges to improve profitability.
• Constance has a big overhaul of the way payout thresholds are set and managed in the website almost complete. Look for that within a day or two. You're definitely going to be impressed.
• Today, I'll be spending my time investigating the network connectivity issues that some miners are experiencing with live data. I expect this issue to be more complex than the memory leak, and it probably will take many days to reproduce and figure out, since it is likely a third-party component configuration problem that is causing the problem.

[Aura89]

Would it be possible to kick idle miners after a certain, acceptable, period of time? and then are not allowed to connect for say, 10-30 minutes?

Some people have complained that their miners connect but aren't actually working, just described, and wish for their miners to be automatically kicked so they can go to the backup pool. This, at least in my view, would allow this to happen, and when the pool allows them to connect after the wait period again, hopefully that time they actually submit shares. This would also, theoretically, prevent connections that aren't working from taking a slot (if the server is full) or a worker that could be working.

[spauk]

i don't think miners would go idle if dynamic difficulty worked properly. most of my miners don't work at 65536 difficulty or higher for some reason. but if dynamic difficulty would adjust rather than staying at the same difficulty, it would certainly find some shares.

[Steve Sokolowski]

Would it be possible to kick idle miners after a certain, acceptable, period of time? and then are not allowed to connect for say, 10-30 minutes?

Some people have complained that their miners connect but aren't actually working, just described, and wish for their miners to be automatically kicked so they can go to the backup pool. This, at least in my view, would allow this to happen, and when the pool allows them to connect after the wait period again, hopefully that time they actually submit shares. This would also, theoretically, prevent connections that aren't working from taking a slot (if the server is full) or a worker that could be working.

There is a feature to do this, but we turned it off. The reason is that some miners don't submit shares because they are so difficult that the miners would be determined to be idle. The only way to solve that would be to lower the maximum difficulty, which would then create database performance issues.

[spauk]

so, returning the server to the way it was when it worked properly (for me), with dynamic difficulty and static difficulty settings, would create database performance issues? i find this kind of hard to believe. sorry if i'm a pain in the ass, i just really want to mine here, and have been unable to use most of my miners here for a couple weeks.