Closed ghost closed 2 years ago
I guess I was wrong regarding (2), the /disconnect does seem to happen, just delayed by ~5 seconds on my system, but that is no problem!
Am I also wrong regarding (1)? Is that supposed to work as well?
I ran some tests and can confirm that (2) works. Without any substantial delay on my system. Probably depends on how fast OpenVPN connection is dismantled.
And, I can also confirm that (1) does not work correctly. When the client main window is closed/hidden it fails to receive WM_ENDSESSION message and the process is killed - having no chance to call /disconnect.
I ran some tests and can confirm that (2) works. Without any substantial delay on my system. Probably depends on how fast OpenVPN connection is dismantled.
No, has nothing to do with OpenVPN, is the API call. But as said, it is fine!
And, I can also confirm that (1) does not work correctly. When the client main window is closed/hidden it fails to receive WM_ENDSESSION message and the process is killed - having no chance to call /disconnect.
Is this fixable?
I am not confortable with the current solution. There has to be a reliable way to perform whatever extensive cleanup is required on sign out. What if we had a database to flush and make consistent on user sign out?
/disconnect
is not calledwhen eduVPN is (completely) closed without disconnecting first,
/disconnect
is not calledAs Windows has no means to keep the VPN active without the eduVPN application running, it can clean up after itself when disconnecting.
Would also fix #171