09-24-2015, 06:20 AM
It's the first time I see such a problem, I'm running the Gallery server on a RasPi and access it via webadmin pretty often, and haven't had any such problem yet.
When it happens again that your browser won't connect, can you run a "netstat -na" in the VM? It'd be interesting to see what connections are currently established and what ports are being listened on.
Also, try a local browser inside the VM (if you don't have graphical access, use the "links" text-mode browser for example).
Where I'm aiming with this: An explanation could be that the browser is using keep-alive connections and the Cuberite http parser doesn't understand them correctly, so the connections get stalled. The browser has a limit of per-server connections it's allowed to use, when it reaches the limit and all the connections are "dead", it claims the server unresponsive. The fix that happens after some time is the browser actually closing some of those connections due to timeouts or possibly other causes.
It's far-fetched, but it's the best I've got so far
When it happens again that your browser won't connect, can you run a "netstat -na" in the VM? It'd be interesting to see what connections are currently established and what ports are being listened on.
Also, try a local browser inside the VM (if you don't have graphical access, use the "links" text-mode browser for example).
Where I'm aiming with this: An explanation could be that the browser is using keep-alive connections and the Cuberite http parser doesn't understand them correctly, so the connections get stalled. The browser has a limit of per-server connections it's allowed to use, when it reaches the limit and all the connections are "dead", it claims the server unresponsive. The fix that happens after some time is the browser actually closing some of those connections due to timeouts or possibly other causes.
It's far-fetched, but it's the best I've got so far