Troubleshooting Information
This section describes how to resolve some of the known issues in CDL configuration. Before proceeding with troubleshooting CDL configurations, consider the following:
-
system-id is unique across the site.
-
cluster-id (datastore id) is unique within the site.
-
You cannot change the replica of Index after deploying the CDL. To change the replica:
-
Set the system mode to "shutdown".
-
Change the replica.
-
Set the system mode to "running".
-
-
The replica of Endpoint, Index, Slot and Kafka pods must be less than or equal to the number of k8 nodes with the following label and value:
-
Label: smi.cisco.com/node-type.
-
Value: Value of cdl/node-type. In a multi-node setup, the default value is session.
-
-
The replica of Zookeeper and ETCD pods must be less than or equal to the number of k8 nodes with the following label and value:
-
Label: smi.cisco.com/node-type.
-
Value: oam.
-
-
In a Geo-replication setup (across the sites cluster-id), the replicas, and maps of the Index and Slot pods must be identical. Otherwise data replication may fail.