Operating System: All
Platform: All

Attachment multiple_ribs.txt has been added with description: Karaf Snapshot showing peer not configured

Attachment Config_datastore.txt has been added with description: Config Datastore showing peer is configured

Steps To Reproduce:

1. Configure 3-Node clustering and install BGP extension.

2. Configure protocol defining BGP RIB ID. Make sure the RIB-ID's are different for all 3 nodes using Open Config.

http://CONTROLLER:PORT/restconf/config/openconfig-network-instance:network-instances/network-instance/global-bgp/openconfig-network-instance:protocols

3. Configure Neighbor using the same RIB-ID as configured above for all the 3-node controllers.

http://CONTROLLER:PORT/restconf/config/openconfig-network-instance:network-instances/network-instance/global-bgp/openconfig-network-instance:protocols/protocol/openconfig-policy-types:BGP/bgp-example/bgp/neighbors

Note: There is a chance this problem get triggered. But not always.

Hi guys, please add ODL version and some logs. So we can clarify where might be the issue coming from, if its duplicated but, etc..

Regards,

Attachment karaf.zip has been added with description: Karaf Logs

(In reply to Claudio D. Gasparini from comment #4)
> Hi guys, please add ODL version and some logs. So we can clarify where might
> be the issue coming from, if its duplicated but, etc..
>
>
> Regards,

Hi Claudio,

I have attached the karaf logs. The ODL version that we found this issue is 0.6.3-Boron-SR3. Please let me know if you need any further information.

Hi Ajay, would be better to have some logs with trace.
Since I don't see any Exception or Error on logs.

Regards,

(In reply to Claudio D. Gasparini from comment #7)
> Hi Ajay, would be better to have some logs with trace.
> Since I don't see any Exception or Error on logs.
>
> Regards,

Hi Claudio,

If you see the config data store shows the rib configuration making 10.18.132.202 as the peer but the karaf logs was continuously saying that 10.18.132.202 has not been configured. Check the following snapshot:

##############################################################################
2017-05-12 08:39:15,267 | ERROR | tLoopGroup-10-14 | BGPDocumentedException | 89 - org.opendaylight.bgpcep.bgp-parser-api - 0.6.3.Boron-SR3 | Error = CONNECTION_REJECTED
org.opendaylight.protocol.bgp.parser.BGPDocumentedException: BGP peer with ip: IpAddress [_ipv4Address=Ipv4Address [_value=10.18.132.202]] not configured.
##############################################################################

Shouldn't the BGP RIB respond correctly and establish second session when in reality it has been configured and showing up in the config data store ?

Let me know if you need to know anything else.

Thanks,
Ajay Chhabria

In this use-case the intention was to use 3 nodes in the cluster as 3 separate BGP speakers with their own RIB. For this to work, OpenConfig model which carries the BGP RIB and peer information needs to be configured on a separate shard and replication disabled for the same

Sample config in configuration/initial/modules.conf file:

Sample config in configuration/initial/module-shards.conf file:

{
name = "openconfig_network_instance"
shards = [

]
}