That script restarts openvpn in case of not finding new routes, and in tap mode no routes are added, so it tries to restart, but the tap0 device is busy, and the openvpn tunnel dies.
After removing that parameter from the startup call, it works as before.
I have this problem with 3.017, but the 3.026 works fine, and the startup command and the ip route check script look the same.
When I remove the --route-up param from the startup command it works, so I’m pretty sure is because of that param, but I don’t know why yet, maybe some component is faster on 3.026 and the check succeeds and it fails on 3.017.
Hi again, as now we are working from home, I’ve experienced the same issue on 3.026 firmware. I’ve narrowed it down to the file /usr/bin/ovpn_check_route, if I insert a sleep command before the check of the routing table at line 14, I’m able to connect.
The minimal amount of delay with I was able to connect with is 3 seconds, so I’m pretty sure it’s related to the network latency.
Is there any way to delay that check without altering system files? For instance, with some directive on the openvpn configuration file.