These are merely API hiccups, either caused locally or wide(actual stat server issues), and generally have absolutely no effect on your mining operations. It just means the agent wasn’t able to communicate with the server for these stats at these given times. You can quite easily confirm this by checking pool-side stats and cross-checking the times when these hiccups occur, or, if you have logs enabled, your local logs even.
Thanks for your info. Now it’s not crashing anymore. The problem is the OC settings are too high and now I can run 12 GPUs without crashing… Thanks tokyorave