1) It's only possible to initiate the connection from Awesome Miner to Remote Agent. It's quite common to setup VPN solutions for these scenarios. In the future, it will also be possible to use the Cloud Services to solve these networking scenarios.
2) You will find the CPU / GPU information on the System tab at the bottom of the screen for the selected miner. This information is always available, even when no API is available or the mining process is stopped
3) The rules doesn't support the concepts of GPU load and Internet connection. The GPU load reported by the system isn't always an accurate way to know if the system is performing well.
If you are running with the profit switcher, you can enable "Ignore pools with no accepted shares" in Options dialog, Profit switching section. Then Awesome Miner will automatically change to another pool in case the first isn't performing as it should. The pool will be marked as failed for the number of miners you have configured, and then it will be used again.
You can have a similar setup with the predefined rule "Accept Progres" in Options dialog, Rules section, where you can take action in case the miner isn't making any process on the hashing or the mining pool is down. In this scenario there are however no way of automatically adding back the first pool once available again.
2) Yes I know it's there. But why not to show it on dedicated tab? If I have rigs with supported miners and without them - it would be better to see all needed info without switching tabs.
3) OK. Is there any way to run a simple script, that will return some value and then do some action based on that value?
(For example to run nvidia-smi - get GPU load - calculate average and return it to AM, so that AM could switch/restart the miner.
4) No. I don't need profit switching. But I need to make sure that rigs are working and correctly respond to different troubles automatically.
a) miner crashing. Count number of restarts and if it is greater than 2-3 within 2-3minutes switch to failover mining list.
b) miner hanging and doing nothing. Get GPU load and if idle for 2-5minutes switch to failover mining list.
b) GPUs down. If down -> hardreset.
c) pool down.
d) internet down.
etc.