When an Exchange Administrator creates additional Virtual Directories for Outlook on the Web or Exchange Control Panel, we have found that a few Managed Availability probes fail on Exchange Server 2013 or above. This occurs because the monitoring probes use the Server's FQDN to access the RPC or MAPI/HTTP frontend virtual directory (e.g. https://ex01.contoso.local/rpc/..., https://ex01.contoso.local/mapi/...). Unfortunately when the probes resolve the local machine's FQDN, there is a chance of getting an IP address bound to the manually created secondary website. When this happens, the probe request fails with HTTP Status code 404 (Not Found) because there's no such virtual directory defined in the web site. If the probes keep failing, their corresponding responders are alerted, resulting in execution of a recovery action. In certain instances, the probes will recycle the frontend MAPI/HTTP application pool. Obviously, this is unnecessary.
To avoid this issue, we recommend you use Add-GlobalMonitoringOverride or Add-ServerMonitoringOverride to disable impacted probes.