bgp doesn't reconnect?

RouterOS Forwarding Protocols
1

helo
i have 2 mikrotics:
board-name: CCR1036-8G-2S+ version: 6.28
board-name: CCR1036-8G-2S+ version: 6.31
my problem is: when bgp with my providers disconnect it stays disconnected until i disable and enable peer

if provider re-enables me at his side only – it does’nt help

sep/24 20:34:57 route,bgp,error HoldTimer expired
sep/24 20:34:57 route,bgp,error RemoteAddress=x.x.19.1
sep/24 20:35:53 route,bgp,info TCP connection established
sep/24 20:35:53 route,bgp,info RemoteAddress=x.x.19.1
sep/24 20:35:53 route,bgp,info Connection terminated
sep/24 20:35:53 route,bgp,info RemoteAddress=x.x.19.1
sep/25 21:06:19 system,info bgp peer city1 changed by root
sep/25 21:06:21 system,info bgp peer city1 changed by root
sep/25 21:06:23 route,bgp,info TCP connection established
sep/25 21:06:23 route,bgp,info RemoteAddress=x.x.19.1

my config is simple:

/routing bgp instance
set default as=xxxx

/routing bgp network
add network=x.x.184.0/21 synchronize=no
/routing bgp peer
add in-filter=defolt name=city1 out-filter=our remote-address=x.x.19.1 remote-as=xxxx
add in-filter=defolt name=city2 out-filter=our remote-address=x.x.19.254 remote-as=xxxx

thank you

2

Try 6.30.4 bugfix on both. We have found it to be very stable on BGP.

3

i’ve upgraded to 6.31, but saw the same disconnect after in, disabling\enabling peer helped

4

6.31 is not considered the most stable as it has lots of new features.

We do an enormous amount of BGP consulting with MikroTik and have found 6.30.4 to be very stable peering to multiple providers.

If you have the same issue with your routers after moving to 6.30.4 on two different CCRs to two completely different upstreams, then you might start checking things inside your network at Layer 1/2.

5

thank you, layer 2 is ok because we are receiving about 1G of traffic via it

i’m thinking of telnet script which will check and re-enable peers :slight_smile:

6

There is a small possibility that this is due to particular type of setup your upstream may have on their BGP..

We have seen this type of behavior with e.g. Hibernia networks. When a bgp session goes down under certain conditions, they will not allow a re-establish, until, unless the session is reset (enable / disable peer), and under certain conditions, they actually have to reset the session from their side.