NSX-ALB VIP health-check failure with "Overlapping routes found" when using ClusterIP mode in TKG
Article ID: 372626
Updated On:
Products
VMware Tanzu Kubernetes Grid
VMware NSX Advanced Load Balancer
Issue/Introduction
- VIP access fails because AVI health check has already failed
- When switching to NodePort access, it works well
- Static Route entries seem to exist in ALB controller UI
- AKO pod log says "
error: Overlapping routes found"
# kubectl -n avi-system logs ako-0
2024-07-08T08:23:03.451Z WARNm rest/rest_operation.go:304 key: admin/global, msg: RestOp method PATCH path /api/vrfcontext/vrfcontext-d19e04cc-a0bf-45e6-9557-f579ff6494f5 tenant admin Obj {"static_routes":[{"next_hop":{"addr":"192.168.13.11","type":"V4"}...},.... returned err {"code":0,"message":"map[error:Overlapping routes found]","Verb":"PATCH","Url":"https://####//api/vrfcontext/vrfcontext-d19e04cc-a0bf-45e6-9557-f579ff6494f5","HttpStatusCode":400} with response null- ALB controller UI shows "Failure: Overlapping routes found"
Environment
- All TKG version
- Using NSX-ALB as ClusterIP mode (Dedicated AVI-SE per k8s cluster)
Cause
AKO can't register a new route against the ALB controller because old routing entries already exist with the wrong label. As a result, AVI-SE can't install the correct routing information from the ALB controller.
- ALB controller UI doesn't show label information, so the user can't confirm whether it's a valid routing entry or not
- You can only see the Static Route entries: (ALB controller UI --> Infrastructure --> Cloud Resources --> VRF Context --> global --> Edit --> Static Route)
- "Additional Information" section in this KB explains how to check the routing entry with a label via CLI, not ALB controller UI
Resolution
- Clean up the old routing entries based on k8s node IP address
- AVI-UI --> Infrastructure --> Cloud Resources --> VRF Context --> Select target VRF (default:global) --> Edit --> Static Route
- See "Next Hop" which corresponded with k8s node IP address
- Click the Trash icon against the target routing entry
- Click "SAVE"
- Restart the AKO pod, after that AKO will recreate fresh routing entries
kubectl -n avi-system delete pod ako-0
kubectl -n avi-system get pods # Status: running
kubectl -n avi-system logs ako-0 # Check no ERROR - Check VIP access
- VirtualService health-check will be all green
- Expected routing entries are installed into AVI-SE
- VIP access works well
Additional Information
Troubleshooting Steps
Check the target node IP address
kubectl get nodes -owide
#> NAME STATUS INTERNAL-IP
#> workload-cluster-A-75hv8-4qpvk Ready 192.168.13.11
#> workload-cluster-A-md-0-92b72-7c67d684c-57rhn Ready 192.168.13.47
#> workload-cluster-A-md-0-92b72-7c67d684c-jtrms Ready 192.168.13.51
#> workload-cluster-A-md-0-92b72-7c67d684c-nmzmp Ready 192.168.13.56
Check the current AKO configuration
kubectl -n avi-system get cm avi-k8s-config -oyaml | grep -E 'cloudName:|clusterName:|cniPlugin:|controllerVersion:|serviceEngineGroupName:|serviceType:'
#> cloudName: Default-Cloud
#> clusterName: workload-cluster-A
#> cniPlugin: antrea
#> controllerVersion: 22.1.2
#> serviceEngineGroupName: test-seg
#> serviceType: ClusterIP
Check the routing table in the AVI-SE
ssh admin@$AVI_SE_IPADDR
sudo ip netns list
#> avi_ns1 (id: 0)
#> avi_poll_ns1
#> avi_poll_ns2
sudo ip netns exec avi_ns1 ip route
#> ###.##.0.0/24 via 192.168.13.11 dev avi_eth8 metric 30000 (<--- Required but Not exist during the trouble)
#> ###.##.1.0/24 via 192.168.13.51 dev avi_eth8 metric 30000 (<--- Required but Not exist during the trouble)
#> ###.##.2.0/24 via 192.168.13.47 dev avi_eth8 metric 30000 (<--- Required but Not exist during the trouble)
#> ###.##.3.0/24 via 192.168.13.56 dev avi_eth8 metric 30000 (<--- Required but Not exist during the trouble)
Check the routing information in the ALB-controller
ssh admin@$ALB_CONTROLLER_IPADDR
admin@avi-controller:~$ shell
Login: admin
Password:
[admin:avi-controller]: > show vrfcontext global
Multiple objects found for this query.
[0]: vrfcontext-f790b561-e7f2-49ee-b08c-425aac54e286#global in tenant admin, Cloud Default-Cloud
[1]: vrfcontext-69cacb9b-c92f-4432-b038-57a4ce1196d2#global in tenant admin, Cloud TEST-CLOUD
Select one: 0 # <------ Choose the target Cloud number
+------------------+-------------------------------------------------+
| Field | Value |
+------------------+-------------------------------------------------+
| uuid | vrfcontext-f790b561-e7f2-49ee-b08c-425aac54e286 |
| name | global |
| static_routes[1] | |
| prefix | ###.##.0.0/24 |
| next_hop | 192.168.13.11 |
| route_id | workload-cluster-A-1 |
| labels[1] | |
| key | clustername |
| value | workload-cluster-A | # CHECK: Target cluster name corresponds with "clusterName"?
| static_routes[2] | |
| prefix | ###.##.2.0/24 |
| next_hop | 192.168.13.47 | # CHECK: Is there any duplicated next_hop?
| route_id | workload-cluster-A-2 |
| labels[1] | |
| key | clustername |
| value | workload-cluster-A |
| static_routes[3] | |
| prefix | ###.##.1.0/24 |
| next_hop | 192.168.13.51 |
| route_id | workload-cluster-A-3 |
| labels[1] | |
| key | clustername |
| value | workload-cluster-A |
| static_routes[4] | |
| prefix | ###.##.3.0/24 |
| next_hop | 192.168.13.56 |
| route_id | workload-cluster-A-4 |
| labels[1] | |
| key | clustername |
| value | workload-cluster-A |
| system_default | True |
| lldp_enable | True |
| tenant_ref | admin |
| cloud_ref | Default-Cloud |
+------------------+-------------------------------------------------+Feedback
Yes
No
Powered by
