Yeah, we're aware of it. This is also what phzi was talking about a couple of posts back. The issue is with the stratum protocol, which currently doesn't support changing the extranonce1. Therefore currently the only option is to disconnect a rig in reconnect it to new order. We'll see if we'll find a technical solution for this.
It's possible to fix =). Multi-pools have had this issue in the past.
However, another, organizational solution is already being implemented - will make sure orders don't switch too fast. There will be a minimum order time frame (not too long though, allowing for the buyers still to be flexible) and the possibility for order hash limit. This will allow more concurrent orders, less order switching and will make hash power providers to stay on single order for longer time. Will not be perfect, there will still be some disconnects form time to time. But at the end of the day we have to make some trade-offs between optimizations for sellers&buyers (we want to make a ~100% fair game for sellers&buyers). Stay tuned!
Sounds like a good plan.
---
I was just replying to your comment that everybody should switch from cgminer to sgminer - that's valid for GPUs only.
bfgminer has scaleability issues on Windows. Can't run more than a couple of instances, can't run more than ~20 miners. But that's another story, it's not a problem with nicehash.
I said everyone should stop using outdated mining software, and gave an example in sgminer. Yes, sgminer is opencl only. bfgminer would be an appropriate solution for miners with other hardware. I have no doubt that there are scalability issues on Windows... I also have no idea why anyone would run windows on a dedicated mining rig. And if you're using gridseeds, then a Pi is probably a better host platform solution.
---
Looks like the stratum is completely offline right now? - my miners all just fell back to a failover pool.
Nicehash.com not loading now either.