Interpreting job queue types
EmailEngine uses queues to process background tasks, and there is a lot of these. The exact queue library used is BullMQ. EmailEngine includes a UI for BullMQ called Arena to give a better overview and allow managing these queues. You can access the tool from the Tools->Arena menu.
From Arena, you can see three queue types:
- submit. This queue includes all email sending jobs.
- notify. This queue includes all webhook sending jobs.
- documents. This queue includes indexing information for ElasticSearch.
Each queue is split into different job types that define the lifecycle of a job. The following applies to the submit queue, but it is mostly the same for other queues as well.
- Waiting. This section includes all jobs that need to be processed right away. These are jobs that were inserted into the queue without the
sendAtproperty, or the
sendAttime has been reached, and thus, the job was moved here from the Delayed section.
- Active. These are jobs that are the ones currently being processed. Jobs from Waiting move here one by one. If one job gets processed, another one gets moved here from Waiting. Depending on the result, processed jobs move to Completed (successful deliveries), Failed (too many retries reached), or Delayed (job failed, but
deliveryAttemptshas not been reached yet).
- Completed. This includes jobs that were successfully completed. All successful deliveries end up here. It is informational only as these jobs are not used anymore for anything. When a sending job is moved here, a
messageSentwebhook is emitted.
- Failed. These are jobs that failed too many times. EmailEngine tried to process these
deliveryAttemptstimes in the Active section and always failed. Failure, in this case, means that the SMTP server did not accept the email for delivery. It does not matter what the exact reason was (network error, wrong password, spam filter triggering, etc.). Once a job ends up here, it is not retried anymore. So it is primarily informational only. You can check the error messages and so on for debugging, but these jobs are not used anymore. When a sending job is moved here, a
messageFailedwebhook is emitted.
- Delayed. This type includes jobs that will be processed in the future. This means two types of jobs – these jobs with the
sendAtvalue set to a future date and jobs that failed in the Active queue, but the
deliveryAttemptscounter has not been reached yet, so a new attempt time was calculated, and the job was moved here. Once the delay time has been reached, these jobs are moved to Waiting. If a job fails in Active section and is moved here, a
messageDeliveryErrorwebhook is emitted.
- Paused. You can pause a queue from the Arena UI. If a queue is paused, all jobs that should go to Waiting end up in the Paused section. Once you hit the "unpause" button, these jobs are moved to Waiting.
- Waiting-Children. This type applies to ElasticSearch indexing. It is not used for sending.
NB! By default, Completed and Failed sections are always empty. To enable these sections, you need to navigate to Configuration -> Service, and set a number for the "How many completed/failed queue entries to keep" input field. Changing this value does not apply to existing jobs, only to new ones.