When a queue grows, record a baseline before clearing or forcing anything.
Check:
- queue depth;
- oldest age;
- arrival rate;
- top destinations;
- top SMTP responses;
- source applications;
- available disk/resources.
If one destination dominates, investigate that route. If every destination is backing up, investigate local DNS, networking, TLS, MTA health or upstream relay.
Avoid mass-forcing retries into a destination that is explicitly throttling you.
After recovery, verify queue drain rate and watch for messages that expired during the incident.