Status as of Wednesday, April 26, 2017
Posted: Wed Apr 26, 2017 8:06 am
Hi, here's a status update:
- Yesterday, we focused on three major issues. First, Chris is completing the restoration of the block explorer data. The system is now at a point where the data can be used as soon as it is coming online, so some coins are already available. Chris think that all explorers will be finished by tomorrow morning at this rate. The reason the database is always just barely behind is because 1TB of data is being written by the explorers.
- Second, we resolved the miner hashrate charts. At the same time, we also resolved a problem that has been with the system for years: when it got behind, hashrates would be calculated with the missing data. Please let us know if the hashrate charts still have problems.
- Third, we discovered a method to significantly improve performance by eliminating coins from mining consideration based on selecting a large coin like Gamecredits and ignoring all coins less profitable than it. In almost all cases, there should be no change to coin assignment. In rare cases, it is possible that the coins assigned to users will be "suboptimal," in that a user might see work restarts more frequently. There were some issues with this release that occurred overnight, which we fixed this morning.
- The statistics we are looking for now are whether there have been rejected shares or slowdowns on the website since 9:00am EDT. If there are none or very few, then we plan to stop with the performance enhancements for now, because that means that performance was the cause and it is acceptable at this point. We'll move on to fixing bugs with other parts of the system. Ideally, we will be able to either today, or by the weekend, make a code freeze against the mining server and not restart it again until new features are added. We'll be analyzing things like rejected shares that occurred after 9:00am to determine whether all of those issues were caused by server overload or not.
- I think we figured out why the live worker statistics keep crashing, and why people using the WAMP API are having so many issues. There was a function sending updates 1000x more frequently than it should have been. The issue has been resolved. Hopefully it will also resolve the problem with the forums running out of memory, too.