docs: record ClientHello layout diagnosis and replacement checks
Co-authored-by: GPT-6 <gpt-6@openai.com>
This commit is contained in:
1 parent
195490af85
commit
780ca71e17
1 file changed
+3
@@ -4,6 +4,9 @@
|
|||||||
|
|
||||||
## Готово
|
## Готово
|
||||||
|
|
||||||
|
- **2026-10-10 transport RCA:** captured packet/strace diagnostics only in disposable containers on the target bridge, fixed destination `142.251.153.4`. TCP establishment succeeds (~18ms); failed passes retransmit ClientHello without server ACK advancement or first TLS response, not an HTTP-body timeout. Full-snapshot, conflict-checked reassembly of 24 connections showed the old frontend seed passes all 22 non-final-SNI layouts and times out on both final-SNI layouts (one TLS1.2, one TLS1.3; zero bytes after SNI). Rustls extension order changes between requests. This strongly establishes observed strategy layout sensitivity, not a proven Rustls/Lua bug or exact network drop location. An independent heavy-tier review confirmed the conclusion/limits.
|
||||||
|
- **2026-10-10 replacement candidate:** `--payload=tls_client_hello --lua-desync=fake:blob=fake_default_tls:tcp_ts=-1000:repeats=1 --lua-desync=multidisorder:pos=1,midsld` passed 24/24 diagnostic connections, including final-SNI in both protocols, then 10 standard checker jobs with 3/3 for TLS1.2 and TLS1.3 (60/60 requests using auto DNS). This is a tested replacement hypothesis, not a universal guarantee. Qualification/evidence rules are unchanged. No working profile/image was switched during this tracing task; target Player API remains OK, old target/Xray unchanged. Temporary trace containers were removed after copying private artifacts; raw captures/reports retained in `/home/agent/checker-transport-trace`, `/home/agent/checker-full-trace`, `/home/agent/checker-alternative-trace`, and `/home/agent/alternative-qualification-*.json/.stderr`. Next implementation step, if approved: prefer this candidate through the normal managed verify/save/apply pipeline instead of weakening checks or modifying TLS libraries/timeouts.
|
||||||
|
|
||||||
- **2026-10-10 remaining-gates follow-up:** `24c29be` adds exact GGC scope comparison to healthy monitor probes and authenticated early-timeout rejection for saved candidates (qualification still requires both TLS protocols 3/3). `b965842`/`cb82436` correct actual upstream diagnostic metadata and validate DNS IP literals. Medium-tier independent review and actual remote-capture Go checks verified genuine one/two-protocol early timeouts authorize rejection, while a passing report does not; malformed/infrastructure diagnostics remain fail-closed. All local Go tests/build/static checks pass. Deployed final image includes `cb82436`.
|
- **2026-10-10 remaining-gates follow-up:** `24c29be` adds exact GGC scope comparison to healthy monitor probes and authenticated early-timeout rejection for saved candidates (qualification still requires both TLS protocols 3/3). `b965842`/`cb82436` correct actual upstream diagnostic metadata and validate DNS IP literals. Medium-tier independent review and actual remote-capture Go checks verified genuine one/two-protocol early timeouts authorize rejection, while a passing report does not; malformed/infrastructure diagnostics remain fail-closed. All local Go tests/build/static checks pass. Deployed final image includes `cb82436`.
|
||||||
- **2026-10-10 runtime gates:** temporary target-only policy 15s/1/30s demonstrated successful saved frontend recheck/apply without a new frontend tune run and real GGC rotation (`rr17---sn-n8v7kn7d.googlevideo.com` ↔ `rr16---sn-n8v7znsz.googlevideo.com`) with newly qualified exact-host candidates. A later real saved rejection entered scan/seed fallback: tune run 9 was recorded FAILED when combined revalidation did not pass, and last-good was restored rather than applying insufficient evidence. Thus successful saved-first and safe fallback entry/restoration are verified; a newly applied frontend from that particular fallback attempt is not claimed. Final target generation 25, active IDs [1,13], both groups committed, Player API OK; only NFQUEUE 300 bound. Restored default policy 5m/3/30m; old target remains running and Xray unchanged.
|
- **2026-10-10 runtime gates:** temporary target-only policy 15s/1/30s demonstrated successful saved frontend recheck/apply without a new frontend tune run and real GGC rotation (`rr17---sn-n8v7kn7d.googlevideo.com` ↔ `rr16---sn-n8v7znsz.googlevideo.com`) with newly qualified exact-host candidates. A later real saved rejection entered scan/seed fallback: tune run 9 was recorded FAILED when combined revalidation did not pass, and last-good was restored rather than applying insufficient evidence. Thus successful saved-first and safe fallback entry/restoration are verified; a newly applied frontend from that particular fallback attempt is not claimed. Final target generation 25, active IDs [1,13], both groups committed, Player API OK; only NFQUEUE 300 bound. Restored default policy 5m/3/30m; old target remains running and Xray unchanged.
|
||||||
- **2026-10-10 full-corpus diagnostic:** clean disposable namespace scan tested 8648 TLS1.2 and 4330 TLS1.3 strategies, found 989/935 working respectively, zero scan errors. Diagnostic scan/check chain did not reach check; recovered 1923 intact candidate rows from captured stdout (one row corrupted by interleaved stderr), then independently checked shortlist: 20 passing scenarios, 10 per protocol, each 3/3. Current selector's combined winner passed both protocols 3/3 in two repeats, third repeat had TLS1.2 timeout at 2/3. Raising request timeout from 6s to 20s and fixing a single destination IP did not eliminate sporadic failures; checker Rust client also times out through stable SOCKS. Exact transport cause remains unproven; do not describe timeout as solved or weaken evidence. Safe continuation after validated early failure is implemented; upstream/client-level tracing is the remaining diagnostic work.
|
- **2026-10-10 full-corpus diagnostic:** clean disposable namespace scan tested 8648 TLS1.2 and 4330 TLS1.3 strategies, found 989/935 working respectively, zero scan errors. Diagnostic scan/check chain did not reach check; recovered 1923 intact candidate rows from captured stdout (one row corrupted by interleaved stderr), then independently checked shortlist: 20 passing scenarios, 10 per protocol, each 3/3. Current selector's combined winner passed both protocols 3/3 in two repeats, third repeat had TLS1.2 timeout at 2/3. Raising request timeout from 6s to 20s and fixing a single destination IP did not eliminate sporadic failures; checker Rust client also times out through stable SOCKS. Exact transport cause remains unproven; do not describe timeout as solved or weaken evidence. Safe continuation after validated early failure is implemented; upstream/client-level tracing is the remaining diagnostic work.
|
||||||
|
|||||||
Reference in new issue
Block a user