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
Yes, I also tested the full tunnel with docker and other methods on version 1.9, but the speed is so low that it cannot be used. Do you have to do something special or apply special settings?
@ehsan272727 — v1.9.22 includes the H2-skip-for-tunnel fix (#1040 + #1041) which restored v1.9.14 tunnel performance on Pixel 6 Pro per yyoyoian-pixel's A/B test (0/30 vs 8-10/30 long-poll stalls). "Totally unusable" on Android v1.9.22 specifically is unexpected and worth narrowing.
What I need to debug:
Confirm version: mhrv-rs --version (Android UI shows on main screen).
Speed metric: fast.com through proxy — what Mbps does it show?
Sample log around a slow request: adb logcat -s mhrv-rs from a connected Android dev session, captured during one specific failing browse action.
Specifically check for: 16s, 17s, or 30s patterns in the log — those are the long-poll stall + batch-timeout patterns. v1.9.22 should NOT show those anymore on the tunnel-batch path.
Topology: which GitHub Actions tunnel method are you using? cloudflared Quick / ngrok / cloudflared Named? Or VPS?
If you don't see any 16-30s stall patterns in the log but speeds are slow, that's a different bottleneck (likely Apps Script edge throttle, not h2 perf — see #895 / #1000).
If you DO see the stall patterns on v1.9.22, that's a real regression vs the PR's A/B claim, and we need to dig into why the Android build behaves differently than yyoyoian-pixel's tested Pixel 6 Pro.
Closing this for now because we asked for the version, speed metric, topology, and Android log on May 11, and there has been no follow-up from the original reporter for more than 7 days.
If this still happens on the latest release, please comment here or reopen with:
version:
Android/Desktop:
fast.com or other speed metric:
tunnel-node method:
adb logcat / app log around one slow request:
Activity
madkobs1242 commentedon May 11, 2026
Yes, I also tested the full tunnel with docker and other methods on version 1.9, but the speed is so low that it cannot be used. Do you have to do something special or apply special settings?
therealaleph commentedon May 11, 2026
Reviewed via Anthropic Claude.
@ehsan272727 — v1.9.22 includes the H2-skip-for-tunnel fix (#1040 + #1041) which restored v1.9.14 tunnel performance on Pixel 6 Pro per yyoyoian-pixel's A/B test (0/30 vs 8-10/30 long-poll stalls). "Totally unusable" on Android v1.9.22 specifically is unexpected and worth narrowing.
What I need to debug:
mhrv-rs --version(Android UI shows on main screen).fast.comthrough proxy — what Mbps does it show?adb logcat -s mhrv-rsfrom a connected Android dev session, captured during one specific failing browse action.16s,17s, or30spatterns in the log — those are the long-poll stall + batch-timeout patterns. v1.9.22 should NOT show those anymore on the tunnel-batch path.If you don't see any 16-30s stall patterns in the log but speeds are slow, that's a different bottleneck (likely Apps Script edge throttle, not h2 perf — see #895 / #1000).
If you DO see the stall patterns on v1.9.22, that's a real regression vs the PR's A/B claim, and we need to dig into why the Android build behaves differently than yyoyoian-pixel's tested Pixel 6 Pro.
Leaving open until you can post the log.
therealaleph commentedon May 22, 2026
Closing this for now because we asked for the version, speed metric, topology, and Android log on May 11, and there has been no follow-up from the original reporter for more than 7 days.
If this still happens on the latest release, please comment here or reopen with:
Answered via LLM, Supervised @therealaleph