Here are a list of the bugs we are aware of. First, some minor bugs that can be ignored and will be fixed last. There is no impact on earnings from these issues:
- The concept of "rejected hashrate" is tricky. If a miner submits a huge number of difficulty 1 shares that get rejected, then the rejected hashrate on the earnings page is being computed by using the sum of the target difficulties of the rejected shares. Chris and I realized that the "rejected hashrate" should be the sum of the actual difficulties of the rejected shares, because that more accurately describes the work lost. This number is far less than the target difficulty, so miners are being misled by the apparently huge amount of lost work. We have corrected this calculation but have not yet released the changes because it isn't serious enough to lose time restarting the server. We'll put the change along with the next round of fixes; until then, just concentrate on whether your "accepted hashrate" is correct.
- The list of recent shares requires too long to load from the website, because the number of coins being mined has increased significantly. The only impact of this issue is that you will have to wait a while to see the table until we resolve the problem. Edited: this issue has been resolved and the changes are immediately available.
- The last 5 minutes shares calculations on the charts pages are incorrect, but again the actual earnings are being computed correctly. Edited: this issue has been resolved and the changes are immediately available.
- Some miners are seeing messages that they found blocks in their mining software over and over and over, even though they didn't actually find a block. We aren't yet sure what the impact of this issue is, or if there is any impact other than the miner printing out these messages.
- Kires is having problems with his miners getting stuck unless they are configured only to connect here with no backup pool. Unfortunately, kires is also the only person who seems to have this problem. We may need to ask him to use his miners for testing later tonight.
- Profitability is too low because the previous release was causing us to orphan our own blocks. Now, we reject blocks from miners who submit them after another miner from our pool submitted the block. Orphan rates are calculated over a one-week window, so the rates are too high because of this now-resolved bug. No action is needed here; as the new software simply does not submit blocks that are guaranteed to be orphaned, orphan rates will fall and payouts to miners will rise over the course of the next week. It should be a steady increase as the week goes on.
- Chris said that the pool's reserve is declining extremely slowly, possibly because some miners are being assigned too high a difficulty for coins where the network difficulty is easier than their miners' assigned difficulties. However, the decline could also be caused by not finding litecoin blocks, which is normal. If there is a problem here, the losses are minimal enough that we can work on the problem tonight and fix it without a revert.
Please offer comments on any issues that occurred as a result of the work restart optimization release. In particular, I want to know of the system's effectiveness on profitability for people who have miners that were most affected by the problem. In our stats, we noticed that most miners mining at the pool have short restart times, which isn't surprising given that others would have been turned away until now.