payout issues?

[Steve Sokolowski]

The issue of pool size is one that Chris and I were talking about.

If you notice now, some coins are in an "error" state because they are too difficult to earn blocks regularly. To reduce our risk, we added a feature to never mine coins where one block would be expected less frequently than some amount of time. With dogecoins, we need to pay people out before ever earning a block.

Once auxiliary mining is implemented within the next ten days, Chris is going to go on a highly aggressive marketing campaign to drum up customers. Up until now, we've been allowing people to dribble in, but our algorithm (other than the merge mining) is more profitable than the other pools and Chris should press the advantage. This week, he's dealing with paperwork like determining where to incorporate the business so that, when the 11th rolls around, we will be able to release the next version and go all-out on marketing for several weeks.

[cryptorific]

What's going on with btc payouts? The past two days haven't happened though I've been over the threshold

[Steve Sokolowski]

I'll ask Chris to investigate this issue. Chris, you there? Reply with your response.

My guess would be that Cryptsy put its wallets in "maintenance mode," which essentially means that they are holding our money without authorization. Chris mentioned that that was a problem recently. He's going to call them today; if that doesn't work, then we'll manually add bitcoins to the wallet with our own money so that people get paid.

[Chris Sokolowski]

I ran the payout script and it worked properly, so the coins have been sent to your wallet. I am now going to investigate why it failed this morning.

[Steve Sokolowski]

Hi Chris,

I suggest looking at the Cryptsy withdrawal time and seeing if Cryptsy had processed the withdrawal, and whether it was confirmed, before 1:15am. The script starts a new database transaction, then attempts to add payouts to the database, and if the audit triggers throw an exception, the script aborts and prints an error to the console (or, since you have it > to a file, in the file). Then, the payouts are added to a bitcoin transaction one by one, and they are attempted to be sent all at once. If even one fails, then the money is not spent and a database rollback is executed. If there was even 0.000001 less than the total amount due, the payout script will fail all payouts for a coin.

However, there are separate database transactions for each coin, because there isn't any reason we don't want to get rid of as much money as possible even if a problem occurs with one daemon. Therefore, it's possible that litecoin payouts succeeded but bitcoin payouts triggered the failsafe.

If a coin's payouts failed, you won't find any evidence of the failure in the database because of this atomicity. You'll only find it in the log file, assuming that you didn't overwrite it when you retried the payouts this afternoon.

[Steve Sokolowski]

Also, there is one more issue you might want to consider. The payout algorithm is designed to fail if transaction fees are greater than 0.5%. A lot of spam has been sent recently to advertise products and services. The amount of money in the spam outputs is lower than the fees required to resend them.

Check our wallets to determine if they are receiving this spam. If so, then we will have to add a Mantis bug to manually delete these inputs from our wallet so that they are never spent.