In the Amsterdam version, when running the Robot HC on SDC running in a multi-node Kubernetes deployment, the SDC-DE HC fails with the message "U-EB Cluster is not available"
curl -X GET http://135.205.228.161:30205/sdc2/rest/healthCheck
- Current response:
{
"sdcVersion": "1.1.0",
"siteMode": "unknown",
"componentsInfo": [
{
"healthCheckComponent": "BE",
"healthCheckStatus": "UP",
"version": "1.1.0",
"description": "OK"
},
{
"healthCheckComponent": "TITAN",
"healthCheckStatus": "UP",
"description": "OK"
},
{
"healthCheckComponent": "DE",
"healthCheckStatus": "DOWN",
"description": "U-EB cluster is not available"
},
{
"healthCheckComponent": "CASSANDRA",
"healthCheckStatus": "UP",
"description": "OK"
},
{
"healthCheckComponent": "ON_BOARDING",
"healthCheckStatus": "UP",
"version": "1.1.0",
"description": "OK",
"componentsInfo": [
{
"healthCheckComponent": "ZU",
"healthCheckStatus": "UP",
"version": "0.2.0",
"description": "OK"
},
{
"healthCheckComponent": "BE",
"healthCheckStatus": "UP",
"version": "1.1.0",
"description": "OK"
},
{
"healthCheckComponent": "CAS",
"healthCheckStatus": "UP",
"version": "2.1.17",
"description": "OK"
}
]
}
]
}
This problem has been raised by someone in the onap-discuss mailing list, in this thread: https://www.mail-archive.com/onap-discuss@lists.onap.org/msg08329.html, and the identified root cause for the failure and proposed solution for this error.
Solution is to change the "ueb_url_list" and "fqdn" fields in /dockerdata-nfs/onap/sdc/environments/AUTO.json, where the lines below should show the dmaap service.ns name identifier instead of the external k8s node IPs, since the containers created can reach the internal dmaap service.ns (i.e. dmaap.onap-message-router with service name being "dmaap" and namespace being "onap-message-router") instead.
To solve this, line needs to be changed in https://gerrit.onap.org/r/gitweb?p=oom.git;a=blob;f=kubernetes/config/docker/init/config-init.sh;h=a65a11e89e8d61f21df1b2c11f6576794e14d34e;hb=refs/heads/amsterdam#l264