-
Notifications
You must be signed in to change notification settings - Fork 10
Description
I have been chasing down a bug for a month or two that causes all reported hashrate to crash to zero and stay there. A restart fixes it. It's been happening ever since someone pointed their nerdminer to the pool. The pattern seems to be that the nerdminer's low hashrate triggers a vardiff bug where the pool tries to negotiate the difficulty level too low and triggers a cascading failure, leading to a channel disconnect loop. This somehow impacts all channels and drops hashrate for the entire pool to zero.
I have tried unsuccessfully to fix this bug many times. But, honestly, I haven't dug in and given it my full attention so I'm obviously missing something.
This time there seems to be a different behavior pattern. Two miners seem to be associated in a failure loop now: THE Dr Nickamoto and the nerdminer (with a testnet address username) are both stuck with zero shares. Actually, in the screenshot the nerdminer found a share. This is very rare because the nerdminer hashrate is absolute garbage. But THE Dr Nickamoto is running a bitaxe and should be finding shares.
This seems to indicate that the bug somehow affects multiple mining channels but not necessarily all of them. It's still a pretty severe failure because all impacted miners are unable to submit shares successfully.
