Logging

YOUnite Uses Elastic

YOUnite logs directly to Elastic via the Elastic RESTtful API.

All activity through the MDM RESTful API is logged synchronously. If Elastic is down, calls to the API will fail with forbidden (403) errors. Once Elastic is available again, the API will function again.

All activity in internal services are logged (a?)synchronously. If Elastic is down the activity in internal services will still succeed, but there will be no logging. Future considerations may include changing to AMQ so as to not drop log entries if Elastic is down.

All activity on the the message queue for the router will be logged (a?)synchronously. This may or may not include the payload, depending on configuration.

Per Resource Indices

Each resource's log entries are kept in separate indices created either when the resource is created or when the resource is accessed, depending on the type of resource. Index names always start with "mdm-". This is used for both easy identification and to take advantage of Elastic's configuration ability to forbid index creation on the fly when the first data entry is posted to it. This allows the logging system to be in control of creating indices and mappings as needed.

Resource indices are created with the naming pattern "mdm-RESOURCE-UUID" where RESOURCE is the type of resource and, UUID is the zone's public facing UUID identifier. E.g. mdm-zone-a2aaedc7-591a-4761-8e35-da875b1e6ac5

Per User (AuthIdentity) Alias's

Alias's are created for the purposes of grouping indices together for ease of use in creating templates in Kibana for dashboards and views that are available through the YOUnite web application.

Alisas names are of the pattern mdm-alias-userUUID where userUUID is the public facing UUID identifier of the authIdentity.

What Gets Logged

The following items are currently logged for API access (see the table below for a complete list of logged events and what is logged):

• date/time
• API path
• resource type (matches the index type, e.g. mdm-zone is resource type zone, mdm-domain is resource type domain)
• resource UUID
• whether the access was allowed or rejected via OAuth
• ssoid of the OAuth token presented for access
• operation type (GET, PUT, DELETE, etc.)

The following items are currently logged for the resource service layer, which is internal, used by both the API and the message router, CRUD access to the resource:

• ssoid
• timestamp
• resource type (same as the index type)
• resource UUID
• action (CRUD)

The following items are currently logged for the Message Bus/router layer:

• Data Events
  • API requests made to the /drs endpoint
    • Callbacks generated by POST /drs/uuid/assembler 
  • Data events generated by adaptors
  • Data event routing to adaptors
  • Data event assembling
• Adaptor registration

Either of these current logging entries can be easily extended to log additional request data (the object model being created, updated, etc.) or response data (results object model of a GET, etc.) for complete auditing of not only who made the modification, but what the modification was. This could be done for the entire resource spectrum or for only specific resources. If needed, the payload data could be easily logged to separate index requiring extended permissions for access.

TODO - PROVIDE TABLE OF WHAT IS LOGGED AND WHAT IS LOGGED ON A PER/ZONE BASIS VS ENTIRE ECOSYSTEM

How It Is Secured

Elastic and Kibana are secured via an NGINX proxy that authorizes each request via the OAuth. An OAuth bearer token is carried along on each request via a cookie when the user clicks the application dashboard link, displaying a Kibana-based view or dashboard page. This secures access to Elastic and Kibana from the public internet. If needed, the proxy piece could be integrated with Elastic in a single container to provide tighter security. This was not done in the dev environment for ease of development and testing, but could be done easily for a stand-alone Elastic server(s).

Kibana

When a user clicks a link to display logging information via Kibana, the URL is modified as it is passes through authorization code in the NGINX proxy. The modification uses the information in the OAuth check_token response to pass along the Elastic alias that Kibana will use to populate the dashboard's data. Access to Kibana is allowed if a user has the correct permissions for Kibana access. By default all users have this permission. Logging information that is displayed/available is controlled through the usage of either the alias for the user's UUID, which contains all the indices they have access to, or via Kibana searches hitting specific indices to which the user has direct permissions (also encoded in the OAuth token).

Elastic

Elastic is secured via the same NGINX proxy that protects Kibana. Permissions to allow read access to an index are controlled via the same permissions mechanism in the token used to allow a user's access to the YOUnite API. If a user has GET or ALL permissions to a zone, they have read-only access to the data in the index in Elastic for the same zone.

Logging Entries

The following table is a summary of the what MDM events get logged and what values get logged for each MDM event.

Notifications

For more information goto

Highlevel discussion on what sort of notifications get sent out and how we leverage webhooks.