Route flap after DR goes down

blazej44800 · October 27, 2019, 8:09pm

Hello,

I just experienced different behaviour in OSPF - RouterOS vs. Cisco iOS according to GNS3 simulation.

I have few routers connected to switch. They are running OSPF broadcast network type with DR/BDR election. With RouterOS when DR router goes down, all other routers flush route table for dead-interval and then reinitialize. All these routers have working BDR, why this route flap?

With Cisco iOS in the same topology no route flap happen.

Blažej

CZFan · October 27, 2019, 10:17pm

Are you only experiencing this in gns3 or same on actual Mikrotik devices?

What is your gns3 setup?

blazej44800 · October 28, 2019, 10:02am

I just proved this issue with real devices - simple 4 routers star with switch in the middle.
R1 - PRIO: 200 - 192.168.0.1/24 - LOOPBACK: 192.168.1.1
R2 - PRIO: 190 - 192.168.0.2/24 - LOOPBACK: 192.168.2.1
R3 - PRIO: 180 - 192.168.0.3/24 - LOOPBACK: 192.168.3.1
R4 - PRIO: 170 - 192.168.0.4/24 - LOOPBACK: 192.168.4.1

Tested with default timers (hello 10s, dead 40s) and custom timers (hello 1s, dead 5s).
R1 is DR, R2 BDR. From R4 I’m pinging R2 loopback. Then I disconnect R1 and wait until dead-interval reached. R2 becomes DR, R3 BDR. Attaching screen record: https://youtu.be/jizkX_4a-CI

In GNS3 I had little bigger simulation with the same results. Attaching screenshot.

jprietove · October 28, 2019, 1:44pm

Please provide export of the routers

Enviado desde mi Mi A2 mediante Tapatalk

blazej44800 · October 28, 2019, 3:08pm

Export below.

R1.txt (872 Bytes)
R2.txt (872 Bytes)
R3.txt (872 Bytes)
R4.txt (872 Bytes)

jprietove · October 28, 2019, 3:45pm

Use LOOPBACK ip Address as Router-ID:

R1, /routing ospf instance set [ find default=yes ] router-id=192.168.2.1
The same in all others Routrers

This will solve all your problems

blazej44800 · October 28, 2019, 5:26pm

No change. Did you test it?

jprietove · October 28, 2019, 5:57pm

Sorry, I didn’t… I will give a look and I’ll tell you

jprietove · December 10, 2021, 3:05pm

Well, quite a bit late. I am testing OSPF and BGP routing in new v7 and updating my training documents for v7. So I tried this and yes, routes flaps in v6 AND routes still flaps in v7 after DR goes down.

[Edited]
But, packets are lost in transit for several seconds in the change from BDR to DR. I’m testing it to verify if it is (still) a bug.
[/Edited]
[Edited 2]
Well, after deeper tests, the BUG is still present in OSPF for RouterOS v.7.1. Changing from BDR to DR is not always smooth. Sometimes the routes don’t flap and sometimes they do.
If the routes don’t flap, a continous PING between is broken for 5 seconds. If the routes flap, it takes 10 seconds.
[/Edited 2]