This chapter outlines the concept behind supervision, the primitives offered and their semantics. For details on how that translates into real code, please refer to supervision.
Supervision has changed since classic, for details on classic supervision see Classic Supervision
There are two categories of exception that can happen in an actor:
- Input validation errors, expected exceptions which can be handled with a regular try-catch or other language and standard library tools.
- Unexpected failures, for example a network resource being unavailable, a disk write failing or perhaps a bug in the application logic.
Supervision deals with failures and should be separated from the business logic while validating data and handling of expected exceptions is a vital part of the business logic. Therefore supervision is added to an actor as decoration rather than something that is intermingled with the message processing logic of the actor.
Depending on the nature of the work to be supervised and the nature of the failure, supervision provides the following three strategies:
- Resume the actor, keeping its accumulated internal state
- Restart the actor, clearing out its accumulated internal state, with a potential delay starting again
- Stop the actor permanently
Since actors are part of a hierarchy it can often make sense to propagate the permanent failures upwards, if all children of an actor has stopped unexpectedly it may make sense for the actor itself to restart or stop to get back to a functional state. This can be achieved through a combination of supervision and watching the children to get notified when they terminate. An example of this can be found in Bubble failures up through the hierarchy.
An actor system will during its creation start at least two actors.
This is the top level user provided actor, meant to bootstrap the application by spawning subsystems as children. When the user guardian stops the entire actor system is shut down.
This special guardian has been introduced in order to achieve an orderly shut-down sequence where logging remains active while all normal actors terminate, even though logging itself is implemented using actors. This is realized by having the system guardian watch the user guardian and initiate its own shut-down upon having seen the user guardian stop.
When presented with an actor which failed while processing a certain message, causes for the failure fall into three categories:
- Systematic (i.e. programming) error for the specific message received
- (Transient) failure of some external resource used during processing the message
- Corrupt internal state of the actor
Unless the failure is specifically recognizable, the third cause cannot be ruled out, which leads to the conclusion that the internal state needs to be cleared out. If the supervisor decides that its other children or itself is not affected by the corruption—e.g. because of conscious application of the error kernel pattern—it is therefore best to restart the actor. This is carried out by creating a new instance of the underlying
Behavior class and replacing the failed instance with the fresh one inside the child’s
ActorRef; the ability to do this is one of the reasons for encapsulating actors within special references. The new actor then resumes processing its mailbox, meaning that the restart is not visible outside of the actor itself with the notable exception that the message during which the failure occurred is not re-processed.
Lifecycle Monitoring in Pekko is usually referred to as
In contrast to the special relationship between parent and child described above, each actor may monitor any other actor. Since actors emerge from creation fully alive and restarts are not visible outside of the affected supervisors, the only state change available for monitoring is the transition from alive to dead. Monitoring is thus used to tie one actor to another so that it may react to the other actor’s termination, in contrast to supervision which reacts to failure.
Lifecycle monitoring is implemented using a
Terminated message to be received by the monitoring actor, where the default behavior is to throw a special
DeathPactException if not otherwise handled. In order to start listening for
Terminated messages, invoke
ActorContext.watch(targetActorRef). To stop listening, invoke
ActorContext.unwatch(targetActorRef). One important property is that the message will be delivered irrespective of the order in which the monitoring request and target’s termination occur, i.e. you still get the message even if at the time of registration the target is already dead.
It can happen that while a message is being processed by an actor, that some kind of exception is thrown, e.g. a database exception.
If an exception is thrown while a message is being processed (i.e. taken out of its mailbox and handed over to the current behavior), then this message will be lost. It is important to understand that it is not put back on the mailbox. So if you want to retry processing of a message, you need to deal with it yourself by catching the exception and retry your flow. Make sure that you put a bound on the number of retries since you don’t want a system to livelock (so consuming a lot of cpu cycles without making progress).
If an exception is thrown while a message is being processed, nothing happens to the mailbox. If the actor is restarted, the same mailbox will be there. So all messages on that mailbox will be there as well.
If code within an actor throws an exception, that actor is suspended and the supervision process is started. Depending on the supervisor’s decision the actor is resumed (as if nothing happened), restarted (wiping out its internal state and starting from scratch) or terminated.