Copy from CORS-2068:

With SDE-1447 we will need Hive/installer logs to include some tracing information and information about the cluster resource. We need this for supportability reasons.

Logs need to include the following tracing fields:

• correlation_id (see x-ms-correlation-request-id in docs below)
• client_request_id (see x-ms-client-request-id)
• request_id (see x-ms-request-id)
• client_principal_name (see x-ms-client-principal-name)

This information can be provided to Hive/installer from ARO RP.

More information on these fields can be found here:

• https://github.com/Azure/azure-resource-manager-rpc/blob/f3c907ae1854b5cda7af18fe6ef91a4b90efe7af/v1.0/Addendum.md#instrumentation-and-tracing-across-services
• https://github.com/Azure/azure-resource-manager-rpc/blob/f3c907ae1854b5cda7af18fe6ef91a4b90efe7af/v1.0/common-api-details.md

We also need the fields to identify the Azure resource we are working with:

• resource_id - can be provided by ARO RP - e.g /subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER
• subscription_id - can be provided by ARO RP or parsed from resource ID - $AZURE_SUBSCRIPTION_ID from resource_id
• resource_group - $RESOURCEGROUP from resource_id
• resource_name - $CLUSTER from resource_id

All these fields must be present in logs when Hive works in context of a specific cluster:

• Provisioning DNS zone for a cluster - fields must be present
• Provisioning or deprovisioning cluster - fields must be present
• Any other operations against a cluster - fields must be present

Disussions have begun therein with padillon, including that the installer will need to be given these k/v pairs by hive somehow when invoked. Details to be added as they become available.