As seen in the parent epic and discussed in slack, AWS Route53 limits API requests to 5/sec per AWS account.

We hit this limit for a number of reasons and multiple axis are investigated and worked on.

One of them is external-dns-operator which seems to be calling the API a lot (about 17K calls per hour, mostly on ListHostedZones and ListResourceRecordSets.

Looking at the external-dns AWS throttling guide, there are multiple options which can be used to limit AWS API calls. Some of them are already in place (domain filtering, resource selection). Some others could be optionally added.

For example:

• --interval: the default is 1m.
  • Since we already use --events, this interval option is only meant to reconcile unmanaged drifts. This could probably be set to 5m or 10m easily.
• --aws-zones-cache-duration: the list of zones is not cached by default.
  • We could set this cache duration to 5 to 10min.

Hypershift operator should have new optional parameters to set those (and/or others) external-dns options so we can configure them and limit impact on AWS API.