Event watcher should not quit upon errors/exceptions #161
Comments
bwaidelich
added a commit
to bwaidelich/Neos.EventSourcing
that referenced
this issue
Feb 28, 2018
This centralizes the invokation of `when*()` event handlers to a central
class that ensures the following invokation order:
# `beforeInvokingEventListenerMethod()` (when implementing `ActsBeforeInvokingEventListenerMethodsInterface`)
# `when*()` (the actual event handler)
# `saveHighestAppliedSequenceNumber()` (when implementing `AsynchronousEventListenerInterface`)
# `afterInvokingEventListenerMethod()` (when implementing `ActsAfterInvokingEventListenerMethodsInterface`)
This allows EventListeners to wrap the event handling in transactions for example:
```
class SomeEventListener implements AsynchronousEventListenerInterface, ActsBeforeInvokingEventListenerMethodsInterface, ActsAfterInvokingEventListenerMethodsInterface
{
public function beforeInvokingEventListenerMethod(EventInterface $event): void
{
// start transaction
}
public function whenSomeEvent(SomeEvent $event): void
{
// db interaction
}
public function saveHighestAppliedSequenceNumber(int $sequenceNumber): void
{
// update sequence number in db
}
public function afterInvokingEventListenerMethod(EventInterface $event): void
{
// commit transaction
}
}
```
This change...
* Renames `EventListenerLocator` to `EventListenerManager` – the class
is now also responsible for *invoking* event listeners
* Removes `EventPublisher::catchUp()` and moves the remaining logic to
the `EventCommandController`
* Introduces an `ActsAfterInvokingEventListenerMethodsInterface` marker
interface, that is used similarily to `ActsBeforeInvokingEventListenerMethodsInterface`
and allows to wrap event handling in a transaction
* Introduces a new `EventCantBeAppliedException` that allows for easier handling of
exceptions during event handling (see neos#161)
* Removes an unused `projectionCache` configuration
Related: neos#159
|
Duplicate of #208 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Currently, if the
event:watchcommand encounters an exception it just stops.In order to stabilize the event handling we should consider a more resilient implementation.
For example: The worker could re-try to apply events and/or skip failed ones somehow
The text was updated successfully, but these errors were encountered: