Insights / FORTIGATE
FortiGate security review: what to check, in order
A FortiGate that was configured well three years ago is rarely configured well today. Staff change, vendors open “temporary” rules, and firmware falls behind. This is the order I work through on a FortiGate security review. It is written so your own team can use it, too.
1. Administrative access
Most serious FortiGate compromises start at the management plane, so start there.
-
Where can admins log in from? Management (HTTPS, SSH) should never be reachable from the internet. Check which interfaces allow it:
show system interface | grep -f allowaccessRemove
http,telnetand, on WAN interfaces,httpsandssh. Manage over a dedicated management network or VPN. -
Trusted hosts. Every admin account should have trusted hosts set to your management subnets:
config system admin edit "jsmith" set trusthost1 10.10.10.0 255.255.255.0 next end -
Named accounts and MFA. One account per person, no shared logins, and two-factor (FortiToken or your identity provider) on every one. Rename or disable the default
adminaccount. -
Least privilege. Use admin profiles so help desk staff can’t change policy.
-
Lockout and crypto:
config system global set admin-lockout-threshold 3 set admin-lockout-duration 300 set strong-crypto enable end
2. Firmware and advisories
Check the running FortiOS version against Fortinet’s PSIRT advisories and the recommended release for your model. Note the end-of-support date for both the hardware and the firmware branch. An unpatched FortiGate with an exposed management or VPN interface is the single highest-risk finding on most reviews.
3. Firewall policies
- Any/any rules: policies with
allas source, destination and service. Each needs a documented reason or a replacement. - Unused policies: sort by hit count and last used. Policies with zero hits over a full business cycle are candidates for removal, after you confirm with the owner.
- Shadowed policies: a broad rule above a specific one means the specific one never matches.
- Inbound VIPs: every published service should have a reason, a restricted source where possible, and security profiles applied.
- Comments: every policy should say who asked for it and why. If nobody knows, that’s a finding.
4. Security profiles
A policy without inspection only checks addresses and ports. Confirm that internet-bound and inbound policies have the profiles you are paying for: IPS, antivirus, web filtering, application control and DNS filtering. Check whether SSL inspection is certificate inspection or deep inspection, and document the exemptions.
5. VPN
- IPsec: IKEv2 where possible, strong proposals (AES-GCM, SHA-256 or better, DH group 14 or higher), no aggressive mode with pre-shared keys.
- Remote access: MFA on every user, user groups tied to your directory, and split tunneling decided on purpose rather than by default.
- If you still run SSL VPN, review it closely against current advisories. Fortinet has been moving remote access toward IPsec, so plan for that change.
6. Logging and backups
- Traffic logging enabled on the policies that matter, sent somewhere off the box (FortiAnalyzer, a syslog server or a SIEM).
- Alerts for admin logins, configuration changes and VPN failures.
- Automatic configuration backups, stored off the device, and a tested restore.
What a good report looks like
A review is only useful if it leads to changes. Each finding should have a risk rating, the evidence, and the exact change that fixes it, ordered so the highest-risk items are fixed first. Changes should go in under change control, each with a test and a backout plan.
If you’d like an independent review of your FortiGates, book a scoping call or read about the Firewall Security Review.