Status as of Tuesday, November 4

[Steve Sokolowski]

Here's an update for today:

• Most of the issues with the mail and forums server have been sorted out. Whatever is left will be resolved by Chris this morning.
• There was a bad query run last week, to correct the corruption caused by the concurrency issues, after the concurrency issues themselves had been resolved. It caused the displayed balances to fall below what was actually owed in many cases, usually by a small amount. We were able to identify and fix the problem last night. No further divergence between the payouts and the displayed balances has occurred since then, indicating that the concurrency fixes were successful.
• An issue with daylight savings time prevented data from being reported to PoolPicker for Sunday, and will prevent one day of reporting in March. Given that the consequences are so minor and the risk so high, we will not be fixing that issue.
• We are going to place multiple exchanges on hold for a few days to resolve the issue with high-powered miners trying to connect and receiving rejected shares. There isn't any reason why our site should be special and be rejecting more shares than any other site, so there must be some incompatibility in our code with these miners. Kristof, do you have a URL where we can download that BFGminer fork that causes the problem (which may be related to this issue)?
• I'm also going to work on a way to recover more easily from Cryptsy's errors before we roll out multiple exchanges. I think the solution is to just keep retrying every 10 seconds every time they disconnect. Hopefully, they won't interpret that as a DoS attack and ban us, but I don't know what else to do when their feed constantly disconnects.
• There is an issue we discovered that can cause payouts to be too high in some cases because of how price data is recorded in the database - it's another concurrency problem. While the issue with balances has been resolved, this bug causes people to be credited too much money when the system is under heavy load. We will continue to pay out for this error and deploy a fix over the weekend. A side effect of this problem is the "-100000ms" times that you see on the "coins mined per day" chart, and those will go away once we deploy the fix.
• Towards the end of the month, after we release multiple exchanges and have stabilized the site more, we're going to start talking about the multiple-algorithm switching client that we had always planned last December. The end goal of the site was always to have a pool that has one dedicated part for scrypt, and another part that switches between the most profitable coins on any algorithm for GPU usage. Customer feedback will be necessary before we start working on that.
• The BTC Guild is, apparently, taking bids because it was going to shut down due to "regulations" and "hacking risk." The owner of the BTC Guild has always struck me as someone who is quick to panic and make rash statements without waiting to see what happens, and here, he is apparently getting out of the industry because there may be regulations that may affect him at some point in the future in one state. I wondered if we could integrate the BTC Guild into our business and did some research, only to find out that the Guild now consists of only 5% of the bitcoin network, down by a factor of 8 from earlier. Few people, including me, want to invest in a declining business, and I wonder if the real reason the owner wants to sell is because his business is no longer profitable enough to justify operating.