Connect count as active when actually dead on long run
|Operating system type:
|Operating system version:
This bug exists very long time and look like nobody notice it, since most users not run Client for days, weeks and months. But I do.
It is look like - Client will be open with connection to some servers. You return here and start work with server. Like change directory.
Client write in log something like "retrieving directory content" and not follow your requests anymore.
You need wait end timeout when Client realize there no connection and goes reconnect and command.
Or you can press "interrupt connection" (yes, this button active when there no active connection), say yes. Client goes glitch and show some BS like travel on non sync dir, but after resolve it all goes fine.
As I see this happen because in code exists simple check connected server or no. It is done as 32-bit timestamp of last connection/action vs current timestamp. This timestamp overflowed on some intervals. Which leads to pass check and Client start work like there active connection.
I not sure what exact interval for appear this bug, but possible near 48 hour.