Diagnosing an OFFLINE machine

Same problem here with some servers. Example bellow is a Debian 11. The same machine have a OpenVPN that not stop working for months. Firewall is completely off. Tcpdump reports thats packets are out an in in 9993 and other ports:

09:04:33.062445 eno1  Out IP 192.168.1.10.38062 > 50.7.252.138.9993: UDP, length 137
09:04:33.062452 eno1  Out IP 192.168.1.10.30202 > 50.7.252.138.9993: UDP, length 137
09:04:33.062465 eno1  Out IP 192.168.1.10.9993 > 104.194.8.134.9993: UDP, length 137
09:04:33.062476 eno1  Out IP 192.168.1.10.38062 > 104.194.8.134.9993: UDP, length 137
09:04:33.062483 eno1  Out IP 192.168.1.10.30202 > 104.194.8.134.9993: UDP, length 137
09:04:33.175989 eno2  In  IP 103.195.103.66.9993 > 192.168.1.10.9993: UDP, length 629
09:04:33.255201 eno2  In  IP 104.194.8.134.9993 > 192.168.1.10.38062: UDP, length 629
09:04:33.285850 eno2  In  IP 84.17.53.155.9993 > 192.168.1.10.9993: UDP, length 629
09:04:33.420511 eno2  In  IP 50.7.252.138.9993 > 192.168.1.10.38062: UDP, length 629

But status stay offline and peers shows all RELAYED. Suddenly, they become ONLINE for a while, sometimes RELAYED and some direct. Other machine (an old Debian 7) in the same network stay online not relayed all the time, but I never can ping them.
OpenVPN have a log thats helps, but I can’t find any in ZT.