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
Mobinnet ISP is not typically as aggressive as TCI/مخابرات for the #313 throttle, but it does happen. So either AUTH_KEY mismatch (#1) or quota tear / Iran ISP truncation (#3 / #4) is most likely for you.
2. The _2026-04-28T20:54:22_ timestamp is in the v1.8.x window so you're getting the new soft-detection messages. Good.
3. Your specific symptoms map to:
YouTube live video doesn't play >30s → SABR cliff (لود نشدن ویدیو های یوتیوب بیشتر از 59 ثانیه #300). Architectural; can't be fixed in apps_script mode. Workaround: Full mode + VPS, or accept short-video-only.
Slow Telegram + reconnects → addressed in v1.8.4 (just released ~5 minutes ago) via PR fix(tunnel-node): raise long-poll to 15s, adaptive straggler settle #446 raising tunnel-node long-poll from 5s to 15s. Persistent connections (Telegram XMPP) won't reconnect every 5s anymore. Upgrade to v1.8.4 and Telegram should be much smoother.
Pinterest / GitHub partial functionality → likely combination of CF detection + the response cliff for image-heavy pages.
Adult video sites not working → typically these use CDNs that flag Google IPs aggressively. Also: video streaming hits the SABR-style cliff. Same architectural ceiling.
Windows Update errors → Windows Update uses HTTPS, but the binaries are large (sometimes >500 MB). The 30s response cliff truncates the binary downloads. Same as the IDM download issue (A way to download? #441) — Range-aware splicing in v1.8.x roadmap will help; Full mode + VPS is the immediate fix.
Recommended diagnostic sequence:
Step 1: Upgrade to v1.8.4 (just released — DoH bypass + adaptive coalescing + tunnel-node fix should help several of your symptoms).
Step 2: Direct curl test to disambiguate AUTH_KEY mismatch vs quota/ISP:
Run 5-10 times. If JSON success >80% → not ISP throttle, likely AUTH_KEY mismatch. If timeouts >50% → ISP throttle (#313 / Mobinnet variant).
Step 3: If AUTH_KEY mismatch likely, set DIAGNOSTIC_MODE = true at the top of Code.gs in Apps Script editor + redeploy as new version (not edit existing). Then re-test:
If you now see {"e":"unauthorized"} JSON → confirmed AUTH_KEY mismatch. Align values + redeploy.
If you still see the same decoy HTML → it's NOT AUTH_KEY (one of causes 2-6).
(disable_padding: true saves ~25% bandwidth, which compounds favorably under throttle. The alternate google_ip is one of the 12-entry pool — try a few from DEFAULT_GOOGLE_SNI_POOL in src/domain_fronter.rs if your default fails handshake.)
Step 5: For long-running content (live video, Windows updates, large downloads): only Full mode + VPS resolves these. The 30s response cliff is architectural in apps_script mode. See #310 / #333 for VPS setup.
Report back with:
Result of the curl test (success rate / typical time)
Whether v1.8.4 changes the Telegram experience
Whether DIAGNOSTIC_MODE flip changes the body content
We can narrow further with that data.
[reply via Anthropic Claude | reviewed by @therealaleph]
Activity
w0l4i commentedon Apr 28, 2026
did you tried to redeploy and add new deployment id to client ?
w0l4i commentedon Apr 28, 2026
اگر کانفیگ درست باشد در ورژن 1.8.3 مشکلتون باید حل شده باشه
therealaleph commentedon Apr 28, 2026
@AliMovahednia — your symptom list points strongly at a mix of three issues:
1. The body string is the v1.8.x decoy/placeholder pattern.
Per v1.8.3's diagnostic taxonomy,
The script completed but did not return anythingmatches one of 6 candidate causes:Mobinnet ISP is not typically as aggressive as TCI/مخابرات for the #313 throttle, but it does happen. So either AUTH_KEY mismatch (#1) or quota tear / Iran ISP truncation (#3 / #4) is most likely for you.
2. The
_2026-04-28T20:54:22_timestamp is in the v1.8.x window so you're getting the new soft-detection messages. Good.3. Your specific symptoms map to:
Recommended diagnostic sequence:
Step 1: Upgrade to v1.8.4 (just released — DoH bypass + adaptive coalescing + tunnel-node fix should help several of your symptoms).
Step 2: Direct curl test to disambiguate AUTH_KEY mismatch vs quota/ISP:
Run 5-10 times. If JSON success >80% → not ISP throttle, likely AUTH_KEY mismatch. If timeouts >50% → ISP throttle (#313 / Mobinnet variant).
Step 3: If AUTH_KEY mismatch likely, set
DIAGNOSTIC_MODE = trueat the top of Code.gs in Apps Script editor + redeploy as new version (not edit existing). Then re-test:{"e":"unauthorized"}JSON → confirmed AUTH_KEY mismatch. Align values + redeploy.Step 4: If ISP throttle, in config.json:
{ "disable_padding": true, "google_ip": "172.217.16.46" }(
disable_padding: truesaves ~25% bandwidth, which compounds favorably under throttle. The alternategoogle_ipis one of the 12-entry pool — try a few fromDEFAULT_GOOGLE_SNI_POOLinsrc/domain_fronter.rsif your default fails handshake.)Step 5: For long-running content (live video, Windows updates, large downloads): only Full mode + VPS resolves these. The 30s response cliff is architectural in apps_script mode. See #310 / #333 for VPS setup.
Report back with:
We can narrow further with that data.
[reply via Anthropic Claude | reviewed by @therealaleph]
therealaleph commentedon May 15, 2026
Closing this for now because there has not been a follow-up from the original reporter for more than 7 days.
If this is still happening on the latest release, please comment here or reopen with the current version, mode, relevant config shape, and fresh logs.
Answered via LLM, Supervised @therealaleph