OSPF fast timers bug?

RouterOS General
1

I am using ROS7 already on some devices (CCR2004, NetPower 16P, CRS326-24S+2Q+, CRS328-24P-4S+) with layer 3 hardware offloading and I do notice that the OSPF session just keeps on flapping randomly w/out any real reason in the logs.

Scenario:

  • /30 IP between 2 devices
  • Running OSPF PTP
  • No OSPF authentication
  • Running 7.9 stable
  • Timers set: 1 Hello, 2 Dead/ (some are set at 2 hello, 4 dead)

Timers were set too aggressive because this was our workaround for the lack of BFD.
I’ve had the same config on VYOS and Juniper devices and never had any problems with it. No random OSPF flaps.

I even ran a ping test to the neighbor device until I see OSPF session drop, and there wasn’t any ping drops, yet the OSPF session restarts.

Refer to the attached image. It’s either it just “timeout”, or I get “sequence mismatch”.

This happens very random. No pattern seen. Neighbor device is either ROS6/ROS7 with the same setting, no BFD.
ospf issue.jpg
Anyone experiencing the same problem? And what is your work around and how did you deal with it??

2

I posted a reply on Facebook, but try slightly altering your timers. For example, I have hello at 1s and dead at 4s.

3

is not a bug.
Timers set: MUST to be the same on both/all sites

4

Not necessarily. What makes you say that? There’s no such rule as that.
Timers must be identical on the each ends of the OSPF session, not across all sites/network

5

This issue has subsided.
Setting high dead timers somehow prevents the flaps to happen. it just doesnt make sense, all I see is that Mikrotik is poor at handling this.