Main Features and Functionalities

Origin Systems

An Origin System is the "owner" of a case. It is where the case is created, but it also holds the responsibility to update, close, or delete (if required). Every change made to the case after creation besides closing and deletion is an update.

Each Origin System has the responsibility to create a data model for the CaseDetails and therefore different Origin Systems can have different data models for CaseDetails. It is the responsibility of the Origin System to expose the endpoint for fetching CaseDetails by Consumer Systems. The endpoint should be secured by an access token from KeyCloak. The URL to the endpoint is set in the caseDetailsUri attribute on the Case object.

When publishing the four commands (CreateCaseCommand, DeleteCaseCommand, UpdateCaseCommand, CloseCaseCommand) these commands are pushed into four different queues. All commands in each queue will be processed sequentially. But the four queues will be processed in parallel!

Consumer Systems

Consumer Systems are the subscribers of cases and corresponding data delivered through DIS by the Origin Systems.

The structure of data delivered is dictated by each Origin System depending on the need for available data. Only EGDW.CaseMetadata is static and set by DIS. The content of CaseDetails and how it is mapped is a contract between the Origin System and the Consumer System. This means the Id, Created, and Updated are set by the Origin System. When a case is created and sent to DIS, DIS will save the metadata in a database (DB) and set the Updated attribute to the same value as Created. When the case, later on, is updated by the Origin System, the value in the DIS DB for metadata will be overwritten with the Updated value.

NemJournalisering (NJ)

When fully implemented, DIS will replace NemJournalisering entirely. They differ in functionality as NJ is considered a Data Warehouse System for cases, where DIS handles information and status about cases.

Notifications

When a case is created, updated, closed, or deleted by the Origin System each Consumer System is notified with a push notification, by utilizing the RabbitMQ technology. External consumers are not connecting to the same queues as the internal consumers.

Every client will have separate queue with template queue name: dis_notifications_{CLIENT_NAME}.

Every time case is created / updated / close / deleted message is sent to each client's queue who have permission to see that specific case.

To get notifications about cases you need to connect to Rabbit instance (see the Environments, Endpoints, and Resources section).

To do that you'll need to do proper auth (see more in Authentication part). After connection you can subscribe to your queue.
You will get notifications only about cases that your client have access to.

Persistence

To make sure that notifications will go out before sending notifications DIS is checking on health of external rabbit connection. To achieve this additional consumer is introduced to external rabbit (dis-healthcheck-stub) only for purpose of reporting connection health. If rabbit is down message is redirected to internal queue (dis-DelayedExternalNotifications) along with the list of clients that we should notify. Then consumer of this queue will try to send messages to external rabbit.

Notifications model

The messages that will go into the notifications queues are:

DIS.Queue.Models.IExternalSystemCaseNotification

and data inside:

    public interface IExternalSystemCaseNotification
    {
        Guid Id { get; }
        ExternalNotificationType NotificationType { get; }
        string? CaseDetailsUri { get; }
        DateTime? EventTime { get; }
    }

    public enum ExternalNotificationType
    {
        Created = 0,
        Updated = 1,
        Closed = 2,
        Deleted = 3
    }

How to Integrate

• Obtain KeyCloak credential from the Cross team.
• Connect to RabbitMQ:
  • Obtain credentials for RabbitMQ from EG DW.
  • Connect to RabbitMQ for push notifications (see the Authentication section).

Environments, Endpoints, and Resources

Read about usage for different environments: External Environments for Shared Tech Products.

The following is an overview of what endpoints will return which data. Can also be seen at the Swagger-endpoint.

HTTP Status Codes

If one asks for case details and the response will be HTTP 404 then one should assume that the case has already been deleted and skip processing of that message

Error Handling

Since this is a distributed system it is expected that in some cases one could have some de-synchronized events / data. For example one could be processing CaseCreatedEvent on a case that has already been deleted. If one asks for case details and the response will be HTTP 404 then one should assume that the case has already been deleted and skip processing of that message - that way we'll get eventual consistency.

Testing

For a Consumer System to test notifications from RabbitMQ the system need to subscribe to an Origin System and the Origin System would have to process a case.

To test if the DIS application is reachable use the Canary endpoint.

Privacy and Security

Authentication

To be able to connect to Rabbit instance you're going to need user and password with access to proper queue. Your contact at EG DW will provide them to you.

For API endpoint security KeyCloak is used for auth.

To authorize a consuming system and make calls to DIS an access bearer token is needed. This token is generated using KeyCloak, with the use of clientId and clientSecret. To add a system to KeyCloak a realm manager (the Cross team) has to set up the client. Once the client has been created the clientId and clientSecret should be provided to the external integrating system. The integrating system should implement the call to KeyCloak to obtain the bearer token and place this in the header when sending requests to DIS and the Origin System (for CaseDetails).

FAQ

• If /api/Cases is requested with the filter set to get cases which have been updated from xx to yy will you then receive cases created in the same set period?

• Is there any pagination on /api/Cases?

• If a case is deleted, will it then still show up in the list of cases (/api/Cases)?

Change History