The CA APM cluster failover environment had the following configuration:
An "Unknown" running state will be reported instead of "Up" because the MOM is not able to map the Collector FQDN in CEM to the Collector VIP that it is configured with.
That also causes the Automatic Transaction Discovery (ATD) failure: "No matching EM collector found for <FQDN of EM running Tim Collection Service>".
Add a VIP <-> FQDN mapping for the Collector running the Tim Collection Service to the MOM hosts file (can also repeat for all Collector VIPs) and restart the cluster. The MOM will then be able to map the Collector FQDN in CEM to the Collector VIP that it is configured with.
The Collectors then had "EM Running State" of "Up" and the ATD startup error was resolved.
Compare the symptoms/impact of running states "Down" and "Unknown" on this documentation page: