Join our Newsletter — 33% off our NHI Course
Home› Guides› FortiClient EMS Hardening Guide: Securing the Server, Database…
Guide Network Security

FortiClient EMS Hardening Guide: Securing the Server, Database and Endpoint Trust

PDF for newsletter subscribers · free to read online

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 14 November 2026 17 min read
On this page

FortiClient Endpoint Management Server (EMS) pushes profiles, installers and VPN settings to every managed endpoint, so whoever controls the EMS server or its database controls the endpoint fleet. That makes EMS a software deployment tier, not just another management console, and it deserves the same protection as a domain controller or a configuration management server. This guide covers the customer-side controls: network exposure and ports, administrator access, the SQL Server or PostgreSQL database behind EMS, certificates and endpoint registration, high availability, backups and monitoring. It is written for endpoint, network and infrastructure engineers who run EMS on premises or in their own cloud tenancy, and it complements the FortiGate Hardening Guide, which covers the firewall side of the Security Fabric.

This is an independent guide from NHI Mgmt Group, based on Fortinet's public documentation as of November 2026. It is not sponsored or reviewed by Fortinet. Product features change, so confirm settings against Fortinet's current documentation.

Key takeaways

  • EMS is a deployment tool. It can install software and change security settings on every registered endpoint, which is why ATT&CK tracks this class of system under T1072 Software Deployment Tools.
  • Publish only the endpoint ports. Endpoints need Telemetry (TCP/8013 by default) and downloads (TCP/10443); the administration console on TCP/443 belongs on an internal management network.
  • The database is part of the attack surface. The EMS database login holds sysadmin or db_owner rights, so treat the database server, its login and the shared backup folder as Tier 0 assets.
  • Control who can register. A Telemetry connection key, user verification and invitation-only registration stop unknown devices from joining and pulling profiles.
  • Few Super administrators, all of them watched. Use custom admin roles, lockout, session expiry, trusted hosts and SAML SSO, and forward EMS logs to the SIEM.

What EMS controls and why it is a high-value target

EMS is the central server for FortiClient. It deploys FortiClient installers, assigns endpoint profiles (malware protection, web filter, VPN tunnels, ZTNA), shares endpoint posture and tags with FortiGate through the FortiOS connector, and issues ZTNA certificates through its SCEP service. On Windows it runs on Windows Server with Microsoft SQL Server; EMS 7.4 also runs on Linux (Ubuntu 22.04 and 24.04 LTS, Red Hat Enterprise Linux 9, CentOS Stream 9) with Apache, Redis and PostgreSQL services.

Three properties make it a priority for hardening:

  • Fleet-wide reach: a change to one profile reaches every endpoint assigned to it, including changes that switch protection off. An intruder who can edit profiles has a path to T1685 Disable or Modify Tools at scale.
  • Endpoint-facing listeners: remote and roaming users need to reach Telemetry and download services from the internet, so part of EMS is usually exposed, which brings T1190 Exploit Public-Facing Application into scope.
  • Credentials and keys held: administrator accounts, AD and LDAP bind accounts, the database login, the code signing certificate for installers and the EMS CA used for ZTNA. Each maps to T1078 Valid Accounts if stolen.

The Endpoint Security Guide covers endpoint protection as a discipline; this page is about protecting the management server itself.

Network exposure and the EMS port map

Fortinet's best practice guidance is to install EMS on a dedicated server reached by an FQDN or static IP, open only the ports EMS needs and block all other service requests to the EMS address. The table below sorts the documented EMS 7.4 ports by who should be able to reach them.

PortServiceDirectionWho needs itHardening action
TCP/8013 (default)FortiClient Telemetry (endpoint management)InboundAll managed endpoints, including remote onesPublish if remote endpoints connect directly; require the Telemetry connection key; changing the port in EMS locks out endpoints still on the default, so plan it
TCP/10443 (default)FortiClient downloads, AV allowlist signatures, custom web filter pagesInboundEndpointsPublish only with 8013; keep installer downloads behind the same restrictions
TCP/443Web console and API (Apache)InboundAdministratorsInternal management network or jump host only; never on the internet
TCP/80ACME certificate validationInboundACME service (for example Let's Encrypt)Open only if you use ACME; otherwise keep closed and turn on HTTP to HTTPS redirect for any internal use
TCP/8015FortiOS connector (FortiGate connects to EMS)InboundFortiGate management addressesRestrict source to FortiGate interfaces; set the Management IP and Port to an internal address
TCP/4001, 4002SCEP for ZTNA certificatesInboundEndpoints using ZTNAOpen only where ZTNA certificate enrolment is in use
TCP/8871EMS AD connector and AD serversInboundDomain controllersRestrict to domain controller addresses
TCP/389 or 636LDAP or LDAPS to Active DirectoryOutboundEMS to domain controllersUse LDAPS (636) so bind credentials are encrypted
TCP/8443 (default)Chromebook profile serverInboundManaged ChromebooksOpen only if you manage Chromebooks
TCP/25 (default)SMTP alertsOutboundEMS to mail relayPoint at an internal relay that supports TLS

Practical layout choices:

  • Remote HTTPS access: under System Settings > EMS Settings this setting controls browser administration from other hosts. When it is off, administrators can log in only on the server itself. If you turn it on, filter TCP/443 at the host and network firewall so only the management subnet reaches it.
  • Linux installer flags: the 7.4 quick start example installs with --allowed_hosts '*' and --enable_remote_https. Replace the wildcard with the hostnames and addresses administrators actually use, and enable remote HTTPS only when you need it.
  • Connect to local subnets only: an EMS setting that allows connections only from local subnets. It suits estates where remote endpoints always connect through VPN first.
  • Private address with VPN: Fortinet notes that a private EMS address requires external clients to connect over VPN. That design keeps 8013 and 10443 off the internet entirely, at the cost of losing management of endpoints whose VPN is down.

The Edge Device Security Guide explains why internet-facing management services attract early exploitation across vendors.

Administrator access and roles

EMS ships with five predefined roles that cannot be edited or deleted: Super (full access, the only built-in role with the Administration section), Standard (full endpoint and policy permissions, read-only settings), Endpoint (full endpoint permissions, read-only policy and settings), Read-only and Restricted (no permissions). Custom roles combine Endpoint, Policy and Settings permissions, and EMS hides or disables menu items a role cannot use.

Setting or controlWhereRecommended use
Default admin passwordFirst login wizardSet a long unique password at install; the default admin account always holds the Super role, so keep it as a break-glass account
Super role membershipAdministration > AdministratorsRestrict to the smallest group possible, as Fortinet's best practice states; give helpdesk staff Endpoint or a custom role
Admin Lockout AttemptSystem Settings > EMS SettingsLocks an admin out after failed logins (default three attempts)
Admin Lockout PeriodSystem Settings > EMS SettingsLockout duration in seconds; long enough to defeat password spraying, short enough for support
Expire login session after x minutesAdministration > Admin User SettingsSet a short value; 0 keeps inactive sessions open indefinitely
Disable administrators' accounts when inactive for x daysAdministration > Admin User SettingsDisables dormant admin accounts; a Super administrator can reactivate them
Change password after x daysAdministration > Admin User SettingsApplies to built-in users only (the admin user and local EMS administrators)
Restrict Login to Trusted HostsPer administratorLimit each admin to jump host or management subnet addresses
SAML SSOAdministration > SAML SSORoute admin logins through the corporate identity provider so its MFA and conditional access apply
Enable login bannerSystem Settings > EMS SettingsShows an authorised use notice before login

Two scoping details matter in practice. Domain administrators are scoped at domain level, with full access to a domain or none, so plan custom roles around that boundary. Where several business units share EMS, the multitenancy feature (Manage Multiple Customer Sites) separates sites rather than relying on role design alone. For the wider pattern of administering management planes, see the Endpoint Privilege Management Guide.

Database access and server hardening

On Windows, EMS installs with SQL Server Express by default, which has a 10 GB limit; Fortinet recommends SQL Server Standard or Enterprise above 5,000 endpoints, installed through the CLI with parameters such as SQLServer, SQLUser or WindowsUser, BackupDir and SQLEncryptConnection. The installer documentation states the database login needs CONTROL SERVER plus membership in sysadmin or db_owner, and backup permissions. That is a highly privileged identity, so the controls around it carry the risk.

  • Prefer Windows authentication for remote databases: join the EMS and SQL servers to the same domain and use a dedicated domain service account (granted "Log on as a service") rather than a SQL login, so the password is managed and rotated in one place. See How to audit SQL Server service accounts and logins.
  • Keep the database off user networks: allow the SQL Server port only from EMS nodes and database administrators' jump hosts. If several databases share one SQL Server, each listens on its own port, so filter each explicitly.
  • Encrypt the database connection: the installer exposes SQLEncryptConnection; Fortinet's example sets it to no, so decide the value deliberately and test encryption with your SQL Server certificate.
  • Keep secrets out of shell history: the documented CLI installs and HA examples pass SQL and file share passwords on the command line. Run them from a controlled admin session, clear history afterwards and rotate the passwords once installation is done.
  • Protect the backup share: remote database installs use a backup directory on the EMS server that both Apache and the SQL Server service can write to. Restrict that share to those two service identities and back it up somewhere administrators of EMS cannot delete.
  • Harden the host: Fortinet's server hardening list covers account lockout, no empty passwords, least privilege, a host firewall, avoiding plaintext protocols, removing unneeded software, patching the operating system and physical security. Treat it as the floor and add your Windows or Linux baseline.
  • Patch EMS promptly: upgrade EMS regularly and apply vendor security updates promptly, tracking Fortinet PSIRT advisories for EMS. Fortinet also advises turning off "Keep updated to the latest patch" for server installers to avoid unplanned disruption; patch servers through change control instead.

The Database Security Guide covers SQL Server hardening in depth.

Endpoint trust, registration and certificates

An EMS that accepts any device that knows its address will hand profiles, VPN settings and installers to that device. These EMS settings narrow who can join and prove which server endpoints are talking to.

SettingWhat it doesWhen to use it
FortiClient telemetry connection keyFortiClient must present this key when connecting (no semicolons allowed)Always, especially when 8013 is internet-facing; distribute it through the installer, not email
Enforce User VerificationUsers must log in to verified user accounts to register to EMSEstates with AD or identity provider integration
Enforce invitation-only registrationOnly users with an invitation can register (some or all users)BYOD, contractors and phased rollouts
Webserver certificate and Use Webserver certificate for Endpoint ControlReplaces the default certificate on Apache and endpoint control with one you chooseAlways: use a CA-signed certificate whose CN or SAN carries the EMS FQDN
EMS CA certificate (ZTNA)The CA EMS uses to issue ZTNA client certificates through SCEPZTNA deployments; protect it as a signing key
Sign software packagesDigitally signs Windows FortiClient installers created by or uploaded to EMS with your code signing certificateWhenever EMS builds installers for distribution
Invalid Certificate Action (endpoint profile)How FortiClient reacts to an invalid server certificateSet to Deny or Warn, as Fortinet's best practice recommends; Deny once certificates are clean
Require Password to Disconnect from EMSUsers need a password to disconnect or unregister FortiClientPreferred over Disable Unregister, because it keeps a support path

Fortinet also documents rogue endpoint disconnection, which removes an endpoint and stops it reconnecting, and the Delete timeout, which removes deregistered endpoints after a set number of days. If you manage Chromebooks, replace the default Chromebook service account credentials shipped with EMS with your own (the three values change together) and do not keep the default P12 private key password. Certificate lifetimes are shrinking across the industry; see the Certificate Expiry Outages and 47-Day TLS Certificates Guide and the Code Signing Security Guide for the signing key.

Where identity fits

Most of the credentials that matter in an EMS deployment belong to services, not people: the database login or domain service account, the AD or LDAP bind account used for user and group sync, the file share account used in high availability, the SMTP relay account, the FortiOS connector relationship with each FortiGate, and the code signing and ZTNA CA keys. Give each one an owner, store its secret in a vault rather than in installer scripts, scope it to the minimum (a read-only bind account for LDAP, for example) and rotate it on a schedule and after any administrator leaves. The FortiManager and FortiAnalyzer Hardening Guide covers the matching API identities on the Fortinet management side.

Resilience: high availability and backups

  • Backups: Fortinet recommends backing up the EMS database weekly through Administration > Back up Database, plus regular SQL backups and database redundancy. Keep at least one copy offline or in an account EMS administrators cannot reach, and test a restore on an isolated server each quarter.
  • High availability: EMS supports active-passive HA with two or more nodes sharing a remote SQL Server. The primary runs the database scripts (ScriptDB=1), secondaries install with ScriptDB=0, and failover follows the High Availability Keep Alive Interval (default 10 seconds, with 5 to 30 seconds recommended). Endpoints must use an FQDN with DNS failover. Fortinet describes HA as failover, not disaster recovery, so backups are still required.
  • HA example settings are not a hardening baseline: the documented HA walkthrough creates a SQL login with password policy enforcement disabled and the sysadmin role. Keep password policy on where your SQL Server version allows, and use a domain service account where you can.

Implementation plan

  1. Inventory: record EMS version, platform, database type and location, every listening port, every administrator and role, and every service credential.
  2. Close the console: move TCP/443 to the management network, set trusted hosts on every admin account and confirm external scans see only 8013 and 10443 (or nothing, in a VPN-only design).
  3. Fix administrator access: reduce Super membership, create custom roles, set lockout and session expiry, and move administrators to SAML SSO with MFA from the identity provider. Keep the default admin as a break-glass account with its password in a sealed vault.
  4. Lock down the database: move to Windows authentication with a dedicated service account where possible, filter the SQL port, decide on connection encryption and restrict the backup share.
  5. Tighten registration in a pilot group: add the Telemetry connection key, user verification and, where appropriate, invitation-only registration to a pilot set of endpoints. Confirm enrolment and re-enrolment work before applying them to everyone. Roll back by clearing the setting for the affected group.
  6. Replace certificates: install a CA-signed webserver certificate, switch endpoint control to it, move Invalid Certificate Action from Warn to Deny once pilot endpoints connect cleanly, and turn on installer signing.
  7. Turn on monitoring: enable EMS Alerts and Endpoint Alerts with SMTP, keep log level off Debug outside troubleshooting, and forward EMS and host logs to the SIEM.
  8. Prove recovery: run a backup, restore it on an isolated host and, if you use HA, rehearse a failover.

Detection and measurement

Watch for changes that would weaken many endpoints at once, and for access to the console or database from unexpected places.

  • Admin activity: failed and locked-out logins, logins from outside trusted hosts, new administrators, role changes and use of the default admin account.
  • Profile changes: malware protection, web filter or VPN profile edits, especially ones that disable features or unlock settings, and new or changed installers.
  • Endpoint registration: spikes in new registrations, registrations from unexpected networks, and deregistrations or disconnects that did not come through the helpdesk.
  • Database and host: SQL logins other than the EMS service identity, new SQL logins or role grants, new web server files, new services or scheduled tasks on the EMS host (watch for T1505.003 Web Shell style persistence).
MetricTargetWhy it matters
Administrators holding the Super roleTwo or three named people plus break-glassEach Super account can change every endpoint
Admin accounts without trusted hosts or SSOZeroOne exception reopens the console to password attacks (T1110 Brute Force)
Days from Fortinet EMS advisory to patched EMSWithin your critical patch windowEMS is partly internet-facing
Managed endpoints registered with the connection key and verified users100 percentShows unknown devices cannot join
Days since last successful restore test90 or fewerBackups that have never been restored are untested assumptions

Common failure patterns

  • Console published with the endpoint ports. A single NAT rule that forwards the whole server exposes TCP/443 administration alongside Telemetry. Publish ports individually.
  • Everyone is Super. Helpdesk and desktop teams given the Super role because custom roles were never built, so one phished account can change every profile.
  • Database login left as installed. A sysadmin SQL login with password policy disabled, its password still in the install script on a shared drive.
  • Open registration. No connection key or user verification, so any device that reaches 8013 can register and receive VPN and ZTNA settings.
  • Default or self-signed certificates with endpoints set to accept invalid certificates. Endpoints then cannot tell the real EMS from an impostor on a hostile network.
  • HA without backups. Active-passive failover protects against a node failure, not against a deleted or encrypted database.

Framework mapping

Control areaNIST SP 800-53 Rev 5CIS Controls v8.1ISO/IEC 27001:2022 Annex A
Admin roles and least privilegeAC-65, 65.15, 8.2
Lockout and strong admin authenticationAC-7, IA-268.5
Restricting console and port exposureAC-17, CM-74, 128.20
Encrypted connections and certificatesSC-838.24
Patching EMS and the hostSI-278.8
Logging and reviewAU-688.15, 8.16
Backups and restore testsCP-9118.13

Relevant ATT&CK techniques: T1072 Software Deployment Tools, T1190 Exploit Public-Facing Application, T1078 Valid Accounts, T1110 Brute Force, T1505.003 Web Shell and T1685 Disable or Modify Tools.

Practitioner checklist

  • Run EMS on a dedicated, hardened server reached by an FQDN, with only the required ports open.
  • Keep the TCP/443 console internal; set Remote HTTPS access and the Linux allowed hosts value deliberately.
  • Limit the Super role to a few named people and keep the default admin as break-glass.
  • Set Admin Lockout Attempt and Period, session expiry, inactive account disabling and trusted hosts for every admin.
  • Use SAML SSO with identity provider MFA for administrator logins.
  • Use a dedicated, vaulted service identity for the database, filter the SQL port and restrict the backup share.
  • Require the Telemetry connection key and user verification; use invitation-only registration for BYOD and contractors.
  • Install a CA-signed webserver certificate, sign installers and set Invalid Certificate Action to Deny.
  • Forward EMS and host logs to the SIEM and alert on admin, profile and registration changes.
  • Back up weekly, keep an offline copy, test restores quarterly and track Fortinet PSIRT advisories for EMS.

Standards and references

Related NHI Mgmt Group resources: FortiGate Hardening Guide · FortiManager and FortiAnalyzer Hardening Guide · Endpoint Privilege Management Guide · Edge Device Security Guide · Database Security Guide

PDF for subscribers

Take this guide with you

Every section, table and checklist in one PDF, with links to the sources and related guides. Subscribe to our weekly NHI & AI Identity Journal and we’ll email you a download link.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 14 November 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org

PDF Downloads for Newsletter Subscribers

Get the guide

Enter your details and we’ll email you a download link.

The full guide is free to read online. The PDF is a thank you for subscribing.

Fill in all fields marked * and tick the PDF box to continue.