Status as of Sunday, September 15, 2019

[Steve Sokolowski]

Good morning! Our focus over the past few days was on improving the website and resolving algorithm-specific issues.

• For an unknown period of time, Equihash was paying about 25% less than the expected value. Several customers had reported this problem since the beginning of August, saying that they were earning 20% less than they could earn at other pools. We had devoted about 80 hours into researching the problem with little success.
• A few hours ago, Chris had a breakthrough when he noticed that some exceptions were being printed to a log file that nobody had looked at in months. I resolved the exceptions and we are waiting for enough data to see whether the equihash problem has been resolved. We should be able to run queries tonight to see whether payouts have risen to the expected values.
• On September 12, Chris upgraded geth to the latest version for Ethereum. Unfortunately, it looks like the Ethereum developers removed support for subscribing to block updates, and a new method is required to receive block updates in the latest version. The issue was not discovered until the morning of September 15, when customers reported that miners were being disconnected. The disconnects occurred because the system detected that blocks were not being received from the Ethereum client, and that the blocks being mined were stale. Some miners additionally saw that "no new jobs were received after 10 minutes."
• Unfortunately, this issue is extremely complex and we don't expect to be able to resolve the Ethereum issue for a week. We can't revert to the previous version, because upgrading changed the block database, and downloading the latest blocks will require several days. Supporting the new version requires research, implementation, testing, and a release, which isn't feasible before Friday. For the next week, we disabled Ethereum mining and set Ethereum Classic as the anchor coin.
• Constance has completed the new auto-configuration tool and profit calculator at https://prohashing.com/tools. You can use the calculator to review what your profit would be expected using any of the listed miners. If there are missing miners, let us know.
• Since we plan to focus on ASIC mining in the future, we will be adding support for the Blake family of algorithms, like Sia, in the coming months. While we will keep the possibility open of adding support for GPU algorithms, our research has concluded that there are far too many of these algorithms to maintain all the coins profitably when one error can cost a huge amount of money, the market capitalization of them is much lower than the ASIC algorithms and more pools are competing for this smaller amount of money, and there already exists client-side software to switch algorithms that renders server-side support unnecessary. We'll continue to reevaluate this decision as time goes on.
• In the coming days, we plan to add a "q=" password argument that will allow miners to specify the type of ASIC being used. This new password argument, when used, will supersede the power settings and make it simpler to set up miners. The use of this argument will also allow us to gather data on the performance of mining equipment, so that we can run queries to see which miners are not achieving their rated hashrate.

[dnprod]

any asic stats you gather with q= should be looked at with a wary eye since many people over (and even under) clock their asic boxes.
i have 3 L3+ running at three different clock rates, depending where they are located in my home or garage (heat issues.)

[Steve Sokolowski]

any asic stats you gather with q= should be looked at with a wary eye since many people over (and even under) clock their asic boxes.
i have 3 L3+ running at three different clock rates, depending where they are located in my home or garage (heat issues.)

Hmmm, for those people I guess we'll have to make it so that the "w=" password argument overrides the equipment argument. That way, it's easy for most people, but those who change the clock speed can override the defaults.