Unresponsive network detection: record connections closed due to
connectivity related errors. Early data indicating that most connections has been closed by the time a network change happens, which is the main motivation for me to track the count for closed connections caused by connectivity related errors. Some connection may get closed early due to connectivity related issues: 1. public reset by the peer post handshake is typically caused by NAT rebinding. 2. QUIC_PACKET_WRITE_ERROR by self is typically early signal indicating network no longer being usable. 3. QUIC_TOO_MANY_RTOS by self is typically caused by silent blackholing. This change also moves metrics logging to QuicConnectivityMonitor. Bug: 1090532 Change-Id: Ic0d05fc788aeb35b49ae853b0b8b652673c16ea2 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2293261Reviewed-by:Ken Rockot <rockot@google.com> Commit-Queue: Zhongyi Shi <zhongyi@chromium.org> Cr-Commit-Position: refs/heads/master@{#788498}
Showing
Please register or sign in to comment