Resolved work restart testing issue

[Steve Sokolowski]

Throughout the past week, some customers have been reporting what appeared to be an error with work restart testing, but which actually turned out to have nothing to do with work restarts. The problem would cause miners to suddenly display errors after running for a long period of time.

The problem has been resolved, and the changes will be available in the next release. The cause of the problem was that, every 60 seconds, the system checked to see whether a miner had not submitted a share in the last 60 seconds - and if none had been received, then that miner was marked in error.

For miners with a high difficulty, it was more likely that no shares would be submitted during that long period of time, so they would often be timed out with this incorrect error. The solution is to only check whether a share had not been submitted in 60 seconds when the miner is being tested, not all of the time.

[kires]

Excellent and than-q!

[kires]

I just noticed that my miners had both failed work restart testing.

[Steve Sokolowski]

Hi Kires,

Did this problem get resolved?

[Some1notu]

Mine gets the work restart problem in spurts. Sometimes I'll see a wall-o-text for restart requests that after a minute or two the miner goes back to mining normally.

[kires]

I bounced BFGMiner, and they went back to normal. Of course, that's the what always resolves an occurrence of the dreaded restart issue.

[Steve Sokolowski]

I think the restart issue happens only when we restart the mining server. If that's the case, I don't think we're going to devote any more time to it. While it is a pain when we make releases, the releases only occur every few weeks.

[kires]

Fair 'nuff. Thanks for looking into it.