You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
SNI discovery from Google IPs with DPI validation #84
It can be helpful to search google services SNI dynamically by using dns.google service and without need for any non google service
i implemented this feature in #83. and tested it both with VPN connection and mhr proxy chained and sometimes randomly i got few SNI but i could not connect to it with neither MCI nor Shatel but some user may be able to use it.
Thanks for the follow-up — the feature shipped in v1.2.6 (merged from PR #83). Your observation about MCI / Shatel sometimes returning nothing useful matches what we've seen: DNS-over-HTTPS against dns.google has varying success depending on the carrier's resolver, and PTR records for Google IPs resolve mostly to 1e100.net infrastructure rather than user-facing hostnames. Some carriers block the DoH request entirely.
A few things worth noting that might improve the hit rate:
PTR coverage is thin. Google runs most of their frontend on anonymous 1e100.net boxes; only some IPs have PTR records pointing at real service names. The SNI rotation pool (DEFAULT_GOOGLE_SNI_POOL in domain_fronter.rs) is a curated list we maintain — issue-driven additions from users reporting SNIs that pass their specific ISP's DPI. That's often more productive than automated discovery for Iranian ISPs specifically.
The TLS-probe validation step is the useful part. Even if PTR discovery finds many names, the validation step tells you which ones actually handshake against your google_ip without DPI interference. That's worth running after adding any new candidate SNIs to sni_hosts.
If you want, the MCI/Shatel SNIs that work for you plus any discovered via scan-sni can be added to DEFAULT_GOOGLE_SNI_POOL by opening a follow-up issue. We're curating those by carrier observation.
Closing this since PR #83 shipped the functionality. Thanks for the contribution!
[reply via Anthropic Claude | reviewed by @therealaleph]
Activity
therealaleph commentedon Apr 23, 2026
Thanks for the follow-up — the feature shipped in v1.2.6 (merged from PR #83). Your observation about MCI / Shatel sometimes returning nothing useful matches what we've seen: DNS-over-HTTPS against
dns.googlehas varying success depending on the carrier's resolver, and PTR records for Google IPs resolve mostly to1e100.netinfrastructure rather than user-facing hostnames. Some carriers block the DoH request entirely.A few things worth noting that might improve the hit rate:
PTR coverage is thin. Google runs most of their frontend on anonymous
1e100.netboxes; only some IPs have PTR records pointing at real service names. The SNI rotation pool (DEFAULT_GOOGLE_SNI_POOLin domain_fronter.rs) is a curated list we maintain — issue-driven additions from users reporting SNIs that pass their specific ISP's DPI. That's often more productive than automated discovery for Iranian ISPs specifically.The TLS-probe validation step is the useful part. Even if PTR discovery finds many names, the validation step tells you which ones actually handshake against your
google_ipwithout DPI interference. That's worth running after adding any new candidate SNIs tosni_hosts.If you want, the MCI/Shatel SNIs that work for you plus any discovered via
scan-snican be added toDEFAULT_GOOGLE_SNI_POOLby opening a follow-up issue. We're curating those by carrier observation.Closing this since PR #83 shipped the functionality. Thanks for the contribution!
[reply via Anthropic Claude | reviewed by @therealaleph]