ITAS Events Services
Overview
Hivedome have created an Eventing Messaging Service which will enable our clients to interact with events that happen inside ITAS. Interaction with Events are via Web Services.
The basic idea is for Client applications to register a ‘mailbox’ with ITAS and then poll the mailbox for new events. ITAS will publish those events to the mailbox as they happen.
Event Service FAQs
Q1. Are the events in real time? Yes, as soon as they happen in ITAS, a message is sent to any mailbox that is registered for that event type. Clients can set up a ‘Long Poll’ so they receive notification as soon as the even happens rather than at the point of making their request.
Q2. Are the events sent in the correct order? Yes, all events are published to the eventing system in the order they happen. If event order is important to you then do not multi-thread polls (requests) to your mailbox. Ordered delivery is only guaranteed for single threaded, single polling applications.
Q3. Can events be guaranteed delivery? Yes, once a client receives notification that an event has happened, it needs to be read. Only when the reader (application) has confirmed the event will it be removed from the queue.
Q4. What type of events are published? Operational notifications are triggered from within ITAS Apps and cover a wide-range of activities. Data Change alerts are self-configurable and can be setup to trigger on Insert and Update activity.
Q5. What happens to events that are not read? Events that are not read, or rather confirmed as read stay in the queue until they have been confirmed as ready. To prevent a huge backlog of events building up mailboxes not subscribed to will time out after 10 day and all messages removed.
API Version History
| v1 | Introduced in 2015 the endpoints provide a PEEK-RECEIVE-CONFIRM model to guarantee sequenced delivery of messages generated through ITAS processes and audit changes |
| v2 | Available as part of the 8.10.4 build the second generation of the API has been migrated to ASP.NET Core and made more efficient by switching to a PEEK-CONFIRM model |