multiple openvpn interfaces restart badly

Steps to reproduce:

1. configure multiple openvpn servers
2. service openvpn restart

Expected result: reconfigured network for each openvpn interface brought up
Actually, the expected result is achieved in less than about 10% of tries. In other 90% only one of interfaces get configured fully.

good run:

daemon.notice netifd: Network device 'mTunUdp' link is down
daemon.notice netifd: Interface 'm_tun_udp' has link connectivity loss
daemon.notice netifd: Interface 'm_tun_udp' is now down
daemon.err netifd[1006]: interface_queue_event(124): Queue hotplug handler for interface 'm_tun_udp', event 'ifdown'
daemon.err netifd[1006]: call_hotplug(100): Call hotplug handler for interface 'm_tun_udp', event 'ifdown' (none)
daemon.err netifd[1006]: device_remove_user(1113): Remove user for device 'mTunUdp', refcount=1
daemon.err netifd[1006]: device_set_present(951): Network device 'mTunUdp' is no longer present
daemon.err netifd[1006]: interface_set_available(487): Interface 'm_tun_udp', available=0
daemon.err netifd[1006]: device_release(731): Release Network device mTunUdp, new active count: 0
daemon.notice netifd: Interface 'm_tun_udp' is disabled
daemon.err netifd[1006]: device_set_present(951): Network device 'mTunTcp' is no longer present
daemon.err netifd[1006]: interface_set_available(487): Interface 'm_tun_tcp', available=0
daemon.err netifd[1006]: interface_queue_event(124): Queue hotplug handler for interface 'm_tun_tcp', event 'ifdown'
daemon.notice netifd: Interface 'm_tun_tcp' is now down
daemon.err netifd[1006]: interface_queue_event(124): Queue hotplug handler for interface 'm_tun_tcp', event 'ifdown'
daemon.err netifd[1006]: device_release(731): Release Network device mTunTcp, new active count: 0
daemon.notice netifd: Interface 'm_tun_tcp' is disabled
daemon.err netifd[1006]: device_remove_user(1113): Remove user for device 'mTunTcp', refcount=1
daemon.err netifd[1006]: task_complete(109): Complete hotplug handler for interface 'm_tun_udp'
daemon.err netifd[1006]: call_hotplug(100): Call hotplug handler for interface 'm_tun_tcp', event 'ifdown' (none)
daemon.err netifd[1006]: device_set_present(951): Network device 'mTunTcp' is now present
daemon.err netifd[1006]: interface_set_available(487): Interface 'm_tun_tcp', available=1
daemon.err netifd[1006]: device_claim(694): Claim Network device mTunTcp, new active count: 1
daemon.notice netifd: Network device 'mTunTcp' link is down
daemon.err netifd[1006]: device_claim(714): claim Network device mTunTcp failed: -1
daemon.notice netifd: Interface 'm_tun_tcp' has link connectivity loss
daemon.err netifd[1006]: task_complete(109): Complete hotplug handler for interface 'm_tun_tcp'

ip rule shows:

...
3:      from all iif mTunUdp lookup ino
4:      from all iif mTunTcp lookup ino
...

bad run:

daemon.err netifd[1006]: device_set_present(951): Network device 'mTunUdp' is now present
daemon.err netifd[1006]: interface_set_available(487): Interface 'm_tun_udp', available=1
daemon.err netifd[1006]: device_claim(694): Claim Network device mTunUdp, new active count: 1
daemon.notice netifd: Interface 'm_tun_udp' is enabled
daemon.notice netifd: Network device 'mTunUdp' link is up
daemon.notice netifd: Interface 'm_tun_udp' has link connectivity
daemon.notice netifd: Interface 'm_tun_udp' is setting up now
daemon.err netifd[1006]: __device_add_user(1037): Add user for device 'mTunUdp', refcount=2
daemon.err netifd[1006]: interface_queue_event(124): Queue hotplug handler for interface 'm_tun_udp', event 'ifup'
daemon.notice netifd: Interface 'm_tun_udp' is now up
daemon.err netifd[1006]: call_hotplug(100): Call hotplug handler for interface 'm_tun_udp', event 'ifup' (mTunUdp)
daemon.notice netifd: Network device 'mTunTcp' link is up
daemon.notice netifd: Interface 'm_tun_tcp' has link connectivity
daemon.err netifd[1006]: task_complete(109): Complete hotplug handler for interface 'm_tun_udp'

as a result configured routes not get applied for mTunTcp

ip rule shows:

...
3:      from all iif mTunUdp lookup ino
...

I have this issue only restarting openvpn manually, after reboot all is configured correctly.

/etc/config/network:

config interface 'm_tun_tcp'
        option device 'mTunTcp'
        option proto 'none'
        option auto '1'

config interface 'm_tun_udp'
        option device 'mTunUdp'
        option proto 'none'
        option auto '1'

config rule
        option in 'm_tun_tcp'
        option lookup 'ino'

config rule
        option in 'm_tun_udp'
        option lookup 'ino'

/etc/config/openvpn:

config openvpn m_tun_tcp
	option enabled 1
	option config /etc/openvpn/server/m-tun-tcp.conf

config openvpn m_tun_udp
	option enabled 1
	option config /etc/openvpn/server/m-tun-udp.conf

Reproduce the failure with the two interfaces in /etc/config/network and the matching /etc/config/openvpn entries, using service openvpn restart and checking ip rule. Start by tracing netifd's device and hotplug event handling shown in the logs; done means both m_tun_tcp and m_tun_udp are fully configured and both routing rules are applied consistently after restart.