I received information from the system technicians that the problem is due to some Gmail addresses to which the system failed to deliver the message.
When a mail server tries to deliver an email that is temporarily rejected by another server (e.g. Gmail), the message is not 'deleted' immediately, but is stored in a retry queue. Exim then automatically retries the delivery attempt at set intervals, every few minutes.
For example, Gmail uses multiple incoming mail servers (five MX records). For each delivery attempt, Exim checks all available servers. If one is unavailable or refuses to accept, it tries the next one. This can effectively translate into multiple consecutive connections to different Gmail servers (in the case of Gmail, this is 5).
If the message is still undeliverable (for example, due to a full recipient's mailbox or a temporary 4.x.x error), the process is repeated every few minutes. This means that dozens of failed delivery attempts can quickly occur in a short period of time if multiple destination servers are checked each time.
The delivery of an email from a bounce email to x@gmail.com was repeatedly rejected and postponed because the recipient's Gmail reported 'over quota', resulting in the error 452-4.2.2 The recipient's inbox is out of storage space.
The server made 7 delivery attempts (connecting to the Gmail server) during this time frame. Since Gmail has multiple servers (specifically 5) and the attempt was repeated for each one, 40 failed mailbox full errors (452-4.2.2) were recorded. If there are more such recipients, then the number of failed connection attempts increases quickly, leading to the error:
Domain smtp.x has exceeded the max defers and failures per hour (100/100 (100%)) allowed. Message discarded.
