SpitfireLoad Testing & APMWhat's new

Target domains (target verification)

On a shared (SaaS) installation a run sends full load only to domains your workspace has proved it owns, so nobody can send load to someone else's site. On on-premises installations the feature is off and the page is not in the menu.

What happens to an unverified target

The run is not refused, it is scaled down: by default to at most 50 VUs and 2 minutes. VUs, request rates and stages are lowered proportionally. A yellow warning on the run page and a Target verification: … line in the run's events say what was not verified.

These count as unverified:

  • Hosts no verified domain covers (HTTP, WebSocket, browser steps, connections such as gRPC, Kafka, SQL).
  • Targets given as an IP address; use a domain name.
  • Steps whose target is known only during the run (an address from a data file or a variable), and JMeter, Gatling, Locust and k6 tests.

A verified domain covers its subdomains: when example.com is verified, api.example.com gets full load too.

A test that runs more scenarios at once than the VU limit cannot be scaled down (each scenario needs at least one VU); the run does not start and the error is target_unverified. Verify the domains or run fewer scenarios at once.

Verifying a domain

  1. Open Target domains in the menu (adding and deleting domains is for workspace admins).
  2. Type the domain in Add a domain (e.g. example.com) and press Add.
  3. The page gives you a 32-digit token starting with spitfire-verify=. Do one of the two options below; take the values with the copy buttons on the page.
  4. Press Verify. The badge turns Verified; if not, the page shows what was looked for and what was found.

We recommend the DNS option: the record stays in place even when the site moves or is rebuilt.

Option 1: DNS TXT record

Create this record in your DNS panel (your own domain and token instead of the examples):

Field Value
Type TXT
Name _spitfire-verify.example.com (some panels want only _spitfire-verify)
Value spitfire-verify=0123456789abcdef0123456789abcdef
TTL the panel's default

Check:

bash
dig +short TXT _spitfire-verify.example.com

The output must show the token in quotes. If it is empty, wait a few minutes and try again.

Option 2: a file on the site

Publish a plain text file whose only content is the token at https://example.com/.well-known/spitfire-verify. For example in the web server's root folder (your own root instead of /var/www/html):

bash
sudo mkdir -p /var/www/html/.well-known && echo 'spitfire-verify=0123456789abcdef0123456789abcdef' | sudo tee /var/www/html/.well-known/spitfire-verify
curl -s https://example.com/.well-known/spitfire-verify

The last command's output must contain the token. Spitfire asks for the file over https first, then http, and expects 200. Redirects (at most 3) are followed only within the same domain and its subdomains. The check reaches only internet addresses, never the installation's internal network.

How long a proof holds

When a proof is older than 24 hours, it is looked for again when a run against that domain starts. If you removed the record or the file, the domain turns Not verified and the run is scaled down, so leave the record in place.

Installation settings (for admins)

Turned on in the controller with environment variables:

Variable Default Meaning
SPITFIRE_TARGET_VERIFY empty (off) required: target verification on
SPITFIRE_TARGET_VERIFY_MAX_VUS 50 Most VUs against an unverified target
SPITFIRE_TARGET_VERIFY_MAX_DURATION 2m Longest run against an unverified target
SPITFIRE_TARGET_VERIFY_RECHECK 24h How often the proof is looked for again at run start

With Helm (a value separate from networkPolicy.egress):

bash
helm upgrade --install spitfire spitfire/spitfire --reuse-values --set targetVerify.mode=required --set targetVerify.maxVUs=50 --set targetVerify.maxDuration=2m

When it is on, the installation's egress should be public-only too (Installation). Monitors, dry runs, try login and Sweeper scans already run at a small load and are not limited.