Settings

The Settings page under Hub Settings gives cluster administrators a single place to configure options that apply to the entire SmarterMail HA cluster — not just the hub currently logged into. Because these settings are cluster-wide, changing a value here affects every hub and node that's part of the cluster, unlike settings made on an individual node (which only affect that node's own SmarterMail installation). The page is broken into four cards: Server Info, Certificate Management, Failover Timing, and File System.

Server Info

• Cluster Hostname - The primary hostname used to access the SmarterMail HA cluster as a whole. This is similar to the primary hostname assigned to a standalone SmarterMail Professional or Enterprise server, except here it identifies the entire cluster rather than a single machine. This field is required and must be a valid, resolvable hostname (e.g., cluster.example.com), since it's also used when SmarterMail requests and validates SSL certificates for the cluster. Because DNS records and any existing certificates depend on this value, changing the Cluster Hostname after a cluster is already in production should be planned out carefully rather than done on the fly.
• Cluster SMTP Sender Hostname - An optional hostname that overrides the Cluster Hostname specifically for outbound SMTP. If this field is left blank, every hub and node in the cluster falls back to using the Cluster Hostname when identifying itself (for example, in the SMTP HELO/EHLO greeting) during outbound delivery. Setting a dedicated value here lets a cluster administrator rely on a single PTR record for the entire cluster, with each node and standby effectively presenting itself as part of that one registered hostname. This keeps outbound mail resolving consistently for SPF checks, no matter which physical node in the cluster actually sent the message.
  
  Example: If the cluster's PTR record is registered to mail.example.com, setting the Cluster SMTP Sender Hostname to mail.example.com ensures every node's outbound mail identifies consistently under that same hostname, even though the actual sending IP address may belong to any node or standby in the cluster.
• Force all traffic over HTTPS - When enabled, any HTTP request made to the cluster is automatically redirected to HTTPS. This is an easy way to make sure webmail logins, cluster administrator logins, and any other browser traffic can't accidentally be sent unencrypted.
  Note: Turning this on means every domain hosted by the cluster needs a valid SSL certificate installed and bound; otherwise, visitors redirected to HTTPS for a domain without one will hit certificate errors. See SSL Certificates for information on installing and automating certificates.

Certificate Management

This card contains a single field: the password used to encrypt the certificate store folder shared by the cluster's hubs. Setting a password here adds at-rest protection to that folder, and the same password is used whenever SmarterMail needs to read certificates that were added to the folder outside of SmarterMail — for example, a .pfx file dropped in manually rather than generated automatically through the cluster's ACME settings.

Failover Timing

This card lets a cluster administrator control how long the cluster waits before deciding a node has actually failed, versus how long that node itself waits before deciding to shut down. Both values are entered in seconds, must fall between 30 and 300, and must be a multiple of 5.

• Max Time Spent Recovering Node (Seconds) - The amount of time the cluster's leader hub will keep attempting to communicate with a node before giving up and failing that node over to a standby. Default: 60 seconds.
• Node Communication Timeout (Seconds) - The amount of time a node will keep trying to recover its own connection to the cluster before shutting itself down. Default: 30 seconds.

These two timers exist separately because they govern two different decisions. Node Communication Timeout controls when a node, on its own, decides it's lost touch with the cluster and shuts itself down. Max Time Spent Recovering Node controls when the leader hub, watching that same node from the outside, gives up and promotes a standby in its place. Keeping the node's own timeout at or below the hub's recovery window helps ensure a failing node shuts itself down cleanly before a standby is activated in its place, which helps avoid a scenario where both the original node and its standby end up active at the same time.

Example: With the defaults in place, if a node loses its connection to the cluster it will keep trying to reconnect for 30 seconds before shutting itself down. Meanwhile, the leader hub gives that same node up to 60 seconds of silence before it activates a standby (if one is configured and eligible) to take over its domains and users.

File System

• File System Restrictions - A list of file, folder, and user names that SmarterMail HA won't allow to be used (for example, as part of a domain, mailbox, or file name). Clicking into this card opens a dialog listing one restricted name per line. Most of these are reserved device names left over from MS-DOS that Windows still honors today (such as CON, AUX, NUL, PRN, and COM1–COM9), along with a handful of reserved ASP.NET folder and file names (such as App_Data and web.config) that could otherwise collide with SmarterMail's own folder structure.
  Note: These restrictions matter most on Windows Server, where the underlying OS still enforces them. Linux administrators can remove entries from this list as needed, though the default list still ships as part of every SmarterMail HA installation. A Reset button is available in the dialog to restore the default list at any time.