Failure alerts are recorded first and retried until they reach you

Since 13 September PayGlue mails you the moment a purchase ends in a permanent failure. That mail is sent by our background worker, and the worker was missing its mail configuration, so the alert did not go out. We found it on 30 September while looking at a publication whose provider events had failed, and we had not been told either, because an incident was only recorded after a successful send. Both are fixed. The configuration is in place, and the alert now writes the incident down before it tries to mail. If the mail cannot go out, the reason is stored and the nightly job sends it. A publication that has not connected Ghost yet is alerted too.

• The worker can send mail again, and checks for it at every start
• An alert that cannot be sent stays open and is retried every night
• After 48 hours we get one internal notice, also when you could not be reached
• Alerts no longer depend on a Ghost connection being set up