Whilst my equipment is UPS protected there are occasional instances where the equipment still loses power. I've noticed that when the stack powers back up there is no network connectivity through the OpenWRT router.
Digging in a little deeper, it seems the cause is that while the OpenWRT router boots into network ready state in about 20 seconds, my switches (overly complicated Ruckus ICX managed units) take a full 3.5 minutes (!) to boot into a state they can pass traffic.
This seems to confound OpenWRT and leaves the switch-to-router trunk VLAN interfaces in a non-operational state, even after the interfaces eventually come up on the switch.
It only takes a restart of the network on the router to fix this (/etc/init.d/restart) but this manual intervention is preventing the stack from self-recovering.
Obviously I could put a delay into the OpenWRT startup or trigger a scripted network restart after a fixed period of time, but this feels like a band-aid rather than a fix. Any ideas if this can be fixed in the core network config? Is there a setting that will tell it to bring up the interface without waiting for a link?
These are the interfaces that don't pass traffic in this scenario:
config device
option name 'br-lan'
option type 'bridge'
list ports 'eth1'
list ports 'eth5'
list ports 'eth6.10'
list ports 'eth7'
option mtu '9000'
config device
option name 'br-mgmt'
option type 'bridge'
list ports 'eth4.100'
list ports 'eth6.100'
option mtu '9000'
config device
option name 'br-guest'
option type 'bridge'
list ports 'eth4.200'
list ports 'eth6.200'
option mtu '9000'
config device
option name 'br-iot'
option type 'bridge'
list ports 'eth4.300'
list ports 'eth6.300'
option mtu '9000'