We haven't changed anything API-related per se, but maybe you're hitting this thing: there is a very confusing design aspect with the sgminer API that we just had to follow to be compatible. The json keys for e.g. avg hashrate last N secs change when you change the period between prints. We did change the default period from 60 secs to 30 secs at some point, so the API keys have changed as well. The proper (and annoying) way to do it is to first send a "config" command, check the reported "Log Interval" value there. That value, in our case "30" for 30 secs, will then be used in the keys for all other messages, for example the "summary" command will return the key ""KHS 30s". Imho it's an idiotic design, but we had to follow it. We have a command line option for changing the period. If there's anything else you're noticing, let me know.


This looks very strange. We only have one pool connection that is shared by all cards, and the shares we submit are for sure CNv8. I'm guessing the pool might not have updated their error message after the fork? Moreover, in your printout above you have 0.46% rejected shares though, nothing out of the ordinary. I tried the same pool now, no problem for my 550/580/Vega 64 test workstation. In the reported stats every 30 secs, are all four cards hashing ok?