Many site owners assume a subdomain stays hidden as long as nobody shares its name. A developer sets up test.misol.uz, installs an SSL certificate and believes only the team knows the address. In reality, the moment the certificate is issued its name is written to Certificate Transparency logs and becomes visible to the whole world. That is by design, so you need to plan for it.
How subdomains are discovered
Every publicly trusted certificate lands in a CT log together with the names it covers, and services like crt.sh make those logs searchable. One command lists every subdomain that ever received a certificate:
curl -s "https://crt.sh/?q=%25.misol.uz&output=json" \
| jq -r '.[].name_value' | tr 'A-Z' 'a-z' | sort -u
The %25 is a URL-encoded percent sign meaning "any subdomain". Next to www and mail you may see dev, staging, crm or backup. Attackers use this during reconnaissance because it is fast, free and sends no traffic to the target.
Why it matters
A name alone is harmless, but test and internal systems are usually weaker than the main site: old CMS versions, simple passwords, debug output, sometimes a copy of the real customer database. Consider a private school in Tashkent on misol.uz. A developer creates beta.misol.uz to try a new design, loads a full copy of the production database and adds a Let's Encrypt certificate. The admin login uses a weak password. Within days automated scanners arrive, because the name appeared in a CT log, and pupils' and parents' personal data is now at risk.
Wildcard certificates hide the list
With a wildcard certificate for *.misol.uz, only the wildcard name is logged, not each subdomain. Let's Encrypt issues wildcards only through the DNS-01 challenge:
certbot certonly --manual --preferred-challenges dns \
-d "misol.uz" -d "*.misol.uz"
Manual mode does not renew automatically, so for real use pick a plugin for your DNS provider's API. A wildcard only hides the CT entry; a subdomain that exists in DNS can still be found other ways.
Rely on access control, not secrecy
Real protection means restricting who can reach test systems. In Nginx you can combine an IP allowlist with a password:
server {
listen 443 ssl;
server_name beta.misol.uz;
allow 203.0.113.10; # office IP
deny all;
auth_basic "Staging";
auth_basic_user_file /etc/nginx/.htpasswd-beta;
}
Create the password file with htpasswd -c /etc/nginx/.htpasswd-beta team (package apache2-utils or httpd-tools). On shared hosting, use .htaccess or the panel's password-protected directory feature.
Keep real data out of staging
Protections get switched off and forgotten, so staging should never hold real customer data. Replace phone numbers, emails and names with fake values and strip payment data when copying a database. Adding an X-Robots-Tag: noindex header also keeps test sites out of search results. For purely internal services, a private internal CA keeps names out of CT logs entirely, though it needs more administration.
Audit your domain like an attacker
Run the crt.sh query for your own domain today. For each subdomain ask whether the service is still needed, whether it must be reachable from the internet and whether its software is up to date. Delete DNS records you no longer need; in the sayt.uz cabinet this is done on the DNS management page.
Make this check part of every project launch. When a new staging host, CRM or test shop goes live, decide in advance whether it needs a public certificate at all, whether it should sit behind an IP allowlist and password, and who is responsible for removing it when the project ends. Handled this way, CT logs stop being a liability and become a free inventory of everything you have exposed.