I’ve just set up a new GitLab install on an internal network and configured it to use 1&1’s SMTP email servers. I’ve tested sending an email from the 1&1 webmail interface, and from the interactive console using the command Notify.test_email(...).deliver_now.
When I look in the Sidekiq logs in the Admin Area, I see several dead jobs in the mailers queue such as Notify#new_user_email with an error message of Net::SMTPFatalError: 550-Requested action not taken: mailbox unavailable.
The destination address is definitely active (it’s my work email address), and the sending domain is whitelisted in the Exchange server’s spam configuration and I can send emails to the same address both on webmail and via the interactive console.
Any ideas of what to check next and why this only seems to happen in the background job? Is there some other step required to reconfigure sidekiq or something apart from gitlab-ctl reconfigure?