I need to do something about pool management in case of rig failures first, because half of providers either have:
- not permitted save
- saving to different file / working directory
- non-matching default pools on site compared to cgminer startup pools - rendering pool auto removal useless
- smos and auto switching to some xyz pools?
I am thinking about creating special software which would act as a middleman between my site and cgminer. Benefits include:
- no more port forwarding on routers
- ability to merge several cgminer instances together and give customer one single big rig
- forced pools - because this middleman would be running on same machine as cgminer or at least in same network, it would be possible to execute several commands each 10 seconds without worries to overload anything
I understand the issues of some rig providers with current system, however, I have second thoughts on this proposal. The current system in awesome for the serious and professional rig providers. It provides a decentralized system which is not dependent on your system - even if your server/system goes down/is having issues, the currently rented rigs will continue to work without any interruptions. And if someone is having issues it is 100% clear where is the fault is (weather at the rig provider or at the renter, the fault can't be on your side). As soon as you'l introduce whatever kind of proxy (local or remote), you'll loose all this. Any kind of proxy also brings additional latency and overhead and my affect mining performance. And there is also the thing of trust in this "special software" which has to run in my network / on my machines. That said - I welcome the new proposal, but you should definitely
keep also the current system for those rig providers who are willing and able to configure it properly (nat rules, cgminer config, etc.).
And, again, djeZo, just risk a couple of hours and implement the "current price" information ... it really shouldn't be that hard.
The proxy I am talking about here is not running on my server, it would be run by every single person offering rig via leaserig. Think about it as an extension of cgminer - alternative would be to modify cgminer itself and make all leasers use it - but that would bring problems with SHA256 providers, because their software is usually forked from earlier cgminer and there are tons of various ones. The proxy runs on same rig; there is no risk of downtime. And because another software is in the middle, I can change "connect direction". Instead of leaserig connecting to cgminer, I can make cgminer (proxy) connect to leaserig, thus avoiding all NAT issues.
The only 2 issues are:
- coding work is needed
- the proxy must be portable (linux & win) - looks like I cannot avoid good old C here...
There are no additional latencies because local proxies have 0ms latency. The stratum connections would be still ordinary, this is just proxy for cgminers API, nothing more.
The additional benefits I forgot to include are:
- better stats regarding stability
- change reports (this one means we could have true real-time stats of current hashing speed on website - with no 1min checking intervals like now)