Google c2dm transient 401 errors on some AWS instances
Solution 1:
This was fixed by changing the the auth key (an automated process via a URL request to the c2dm service to get the new key, then putting it immediately into our server-side push code). We were reluctant to do this while the key was working fine for most c2dm pushes, but it looks like some Google servers were unhappy with the key, causing intermittent errors for us. We've been error-free since changing it over a week ago.