OSPF drops every 30m

hi,
I’m struggling with weird problem on OSPF point-to-point link over Ubiquit p2p RocketM airlink. This link is keeping disconnect every 30m on OSPF, but keeps connected on LAN and on WLANs.

14:50:23 route,ospf,info OSPFv2 neighbor 10.1.255.218: state change from Full to Init
15:20:23 route,ospf,info OSPFv2 neighbor 10.1.255.218: state change from Full to Init
15:50:23 route,ospf,info OSPFv2 neighbor 10.1.255.218: state change from Full to Init
16:20:23 route,ospf,info OSPFv2 neighbor 10.1.255.218: state change from Full to Init
16:50:29 route,ospf,info OSPFv2 neighbor 10.1.255.218: state change from Full to Init
17:20:29 route,ospf,info OSPFv2 neighbor 10.1.255.218: state change from Full to Init

Interface configuration in OSPF:

/routing ospf interface
add dead-interval=1m hello-interval=5s interface=to_rt218 network-type=point-to-point

Any idea what can be a problem here?
Thanks.

You could try debugging OSPF via logs. If you are sure this isn’t a congestion issue.

You could also try setting up the network type as NBMA (and adding the static NBMA neighbors), UBNT wireless links have (had?) usually issues with multicast traffic (the one OSPF uses), changing to NBMA sets this to use unicast instead.

In any case I’d also make sure OSPF traffic gets prioritized in case of congestion (tricky on wireless, but can be done).

what software release are you on?

Do you have any packet loss between peers?
Also, why did you change default hello-interval and dead-interval? (was 10s and 40s). Same behaviour when using defaults?

IRC OSPF has a panic update every 30 minutes even if there are no changes to the topology.

It’s actually not a “panic update”

It’s part of the standard, the routers that create an LSA needs to reflood said LSA every 30 minutes when there are no changes. However this is a neighbor status change that is managed by the hello packets, INIT state means the router “has seen a hello packet from a potential neighbor”, would be good to check what’s the other side showing in the logs.

I didn’t pay attention to OPs hello and dead timer, why did you change it to such values? 5 seconds hello and 1 minute dead timer, as long these match shouldn’t be any issue (aside for really slow reactions to problems) but I’m curious.

Also did you try my suggestion?

As others have also pointed, you need to be sure this isn’t really a congestion issue as well.

Hello, my first suggestion will be to try pinging each neighbor to discard any issues in that circuit. Is that a fiber or wireless circuit?