I have recently collected data about DNS TimeOuts from these sources:
- Incoming DMARC aggregate reports about my mail when received by other domains
- SPF lookups on incoming mail. (This occurs in a Python module, not in SM). Timeout is 20 seconds.
In both cases, the frequency of DNS Timeout is higher than expected, measured in incidents per 1000 rather than incidents per 100,000 or more.
In DMARC reports, most of reported problems come from "Outlook" or "Enterprise Outlook", and I have already observed that I never know what to believe from them.
My internal data has lower error rates than my aggregate report data.
DNS seems very reliable in every other context.
Any ideas on how to manage this issue?