Had a short server outage for about an hour, this morning.

-=-=-

"Something" happened to the server, though Linode haven't said exactly what, but it was hardware based, and they ended up having to transfer the hard drive to a different backup server "shell" before they could get things booted up again.
Crikey!

Obviously, during this time, things were inaccessible, but one thing that surprised me was that both Browsercade and Shoebox were unplayable during this time.
It took me AGES to get the caching to work well enough that the games would work when the device is offline, but apparently things go completely awry when the SERVER is offline.

Rather than falling back to the cached files, the games would sit there in limbo, trying endlessly to read from a server which "should" be accessible, but isn't.
Neither engines would load anything, and both were just blank screens of nothingness.

Hmm...

I might have to look into ways to bypass that, in future, just in case.
I imagine a Timeout might help..?

Meanwhile..

Speaking of Caching, yesterday I changed the way the main asset folder name works in both Browsercade and Shoebox.
From here on, rather than caching an Asset/ folder, they now cache an Asset_VersionNumber/ folder.
This means that whenever I change images (logos, spritesheets, etc) I can EASILY force a reload of all of the images, instead of relying on the seemingly guess-based caching that devices are doing.

I'd previously attempted adding ?Time=Timestamp to the URLs, but that completely broke the cache.
I could also have renamed each and every file, each and every time, but ... No, that's not ideal!
Simply renaming the folder appears to work well enough. Though I'll have to check that it does indeed flush out the old caches, lest you end up with 80gb of duplicated asset/ folders in your browser!!