Hmm. But refresh, ctrl+refresh, shift+refresh, ctrl+shift+refresh all yield the same result for me: Unable to check for updates
Here's a video, and you can see that using ?id=firmware ends up with a faster result of success, as does manually leaving the page and going back which has ?id=firmware appended:
https://youtu.be/w-BCqqK0BG0
I would think that's indicative of some type of user/connection-induced timeout issue not being at play, if the success is noticeably faster than the failure and successes only occur when the ?id=firmware is appended in this case.
The failures are reproducible in succession of refreshes, multiple times, so whatever would seem to be loading isn't actually loading, and leaving the page and going back will insert ?id=firmware into the URL, which then succeeds at the same approximate speed as when I manually input it. Refreshes fail, and the ?id=firmware is not in the URL.
Since this is an unusual issue, and I'm a magnet for random edge-case scenarios, I'll mention the seemingly-irrelevant situation that this was an original install of 2.6.0, using a backup from an old installation of 2.6.0 which was originally upgraded over time with each version between 2.3.x somewhere, through 2.6.0. This 2.6.0 system was then upgraded to 2.7.0.
Another system with an original installation of 2.7.0 doesn't have this issue, yet the URL doesn't show ?id=firmware when changing branches. Its success is at the same approximate speed as my success when manually using ?id=firmware, and both systems use a similar setup and use the same DNS servers in the same order. I'm using the same browsers for both systems when testing.
Odd, indeed. Would comparing the php pages and php configs be of use just to see what may possibly be different, to rule that out?