🔥 ข่าวด่วน: sushanb/google cloud go/pull/

sushanb · 2026-01-07T05:33:22Z

Notes.

The current metric for directpath eligiblity is async via periodic flush.

In future cl, We will make it synchronous and use direct acces only if

if prime() succeeds within 10s
If synchrous metrics flush for directpath elibility succeeds.

Prior to this change, heartBeatLoop's Timer was armed once at Start to initialHeartbeatGrace (30 min) and only reshuffled after it fired. The atomic nextHeartbeatDeadlineNano was extended by every frame per SESSION_SPEC.md #7, but a later SessionParameters frame that shortened the interval to, say, KeepAlive=100ms would leave the Timer sleeping past its now-relevant deadline — a stalled vRPC in the first 30 min was caught by the caller's ctx, not by the missed-heartbeat watchdog. Add a cap-1 non-blocking heartbeatWake channel; handleSessionParameters and resetHeartbeatDeadline both fire it after updating the atomic. heartBeatLoop's select now watches the wake alongside timer.C, stops and drains the Timer on wake, then re-evaluates against the current atomic. Bursts of wakes coalesce into one pending signal, so hot-path frame handling stays allocation-free. Behavioral impact: missed-heartbeat ForceClose now fires within ~3×interval of the last frame regardless of whether the initial-grace Timer has fired. Retry oracle re-picks a peer session as before (StateTransportFailure + Unavailable, idempotent retry). Follow-ups deferred: the wake fires on every reset, including extension-only resets that the Timer would learn about naturally on next fire. A shrink-only CAS gate is a natural optimization but not a correctness item — see [[bigtable-reactive-heartbeat-followups]].

…Loop, heartBeatLoop) Third and final PR in the Session core stack. Adds the lifecycle orchestration that drives a Session from Start through teardown, plus the readLoop that dispatches server frames to the vRPC handlers landed in bigtable-session-vrpc. Stacks on bigtable-session-vrpc; deletes the minimal ForceClose stub that PR shipped, replacing it with the full lifecycle-shaped body. New file: session_lifecycle.go (~640 LOC) - Session.Start(ctx, OpenSessionRequest) — transitions New→Starting, Sends the OpenSession frame, fires onStart, spawns readLoop + heartBeatLoop. Wraps a failed Send as codes.Unavailable so retry plumbing treats pre-wire OpenSession loss the same as any other transport-side loss. - ForceClose — full body: transitionTo(Closed), setCloseReason, notifyClosing (once), cancelActiveRPCs, signalQuiescent, notifyClosed. - Close(ctx, CloseSessionRequest) — graceful drain: Ready→Closing, waits on quiescent, sends CloseSession, transitions to WaitServerClose, arms the pool's stuck-session sweep to eventually ForceClose if the server never confirms. - notifyClosing / notifyClosed — once-guarded hook dispatchers with the strict onClosing-precedes-onClose ordering enforced. - readLoop(ctx) — Recv-loop that dispatches every SessionResponse variant via handleSessionResponse, drives handleClose on stream termination, records msgsRecv counters + resets heartbeat deadline on every recognized frame. - handleSessionResponse — dispatch switch to handleOpenSession / handleVRPCResponse / handleErrorResponse / handleSessionParameters / handleGoAway / handleSessionRefreshConfig. Heartbeat frames update the deadline via resetHeartbeatDeadline; the unknown-payload branch records a debug tag but doesn't reset the deadline (so a misbehaving server can't keep the watchdog satisfied with junk). - handleOpenSession — Starting→Ready transition, PeerInfo extract from the bidi header (synchronous, matches Java's onHeaders synchrony), fires onActive. - handleGoAway / handleClose / handleErrorResponse / handleSessionParameters / handleSessionRefreshConfig — protocol-level handlers. - heartBeatLoop — Timer + heartbeatWake reactive-wake pair. Only enforces the deadline while a vRPC is in flight (idle sessions legitimately receive no server heartbeats). See SESSION_SPEC.md #7. - peerInfoExtracter — parses the bigtable-peer-info header, stamps s.peerInfo atomically before onActive fires. - closeReasonLabel / closeReasonToCause — CloseSessionRequest.Reason → string / sentinel error mapping for close-reason attribution. New file: session_lifecycle_test.go (~660 LOC) — 30+ tests covering Start, ForceClose, Close, readLoop, handleSessionResponse dispatch, handleOpenSession + peerInfo parsing, handleGoAway, handleClose, handleErrorResponse (rpc_id=0 harmlessly drops via routeVRPCFrame guards), heartBeatLoop (reactive wake, idle-gate, missed-heartbeat ForceClose). Edits to session_vrpc.go - Delete the minimal ForceClose stub introduced in the prior PR — the full body lives in session_lifecycle.go now. Edits to session_test.go - Add hookCounts helper (start/active/close callback counters). - Add setSlotForTest helper (seed the in-flight slot from tests). Stack: - googleapis#20211 — Session debug surface - googleapis#20213 — Session vRPC dispatch + slot lifecycle - This PR — Session lifecycle

Bundles four related changes for the sessionz working branch: 1) 1x heartbeat multiplier (was 3x). resetHeartbeatDeadline and handleSessionParameters now compute `now + interval` instead of `now + 3*interval`. Matches java-bigtable's SessionImpl.java:440 (nextHeartbeat = clock.instant().plus(heartbeatInterval)). Under healthy operation nothing changes — each frame's reset fires before the timer wakes. Only sessions where the server genuinely goes silent for one interval trip the watchdog. 2) heartBeatLoop cleanup: - Delete `active := 1 // multiplex=1; kept as %d so a future >1 stays greppable` code smell (variable existed only for its self-documenting comment). - lastFrameAge = interval - remaining (was 3*interval - remaining) to match the new 1x deadline. - Debugf/recordEvent format strings drop the in_flight=%d field that was carrying the constant 1. 3) Vet fixes matching PR googleapis#20213's CI signal: - virtRpc → virtRPC (initialism convention). - runVRpcCapture and runInvoke: ctx moves to first parameter per Go convention. Updated all 5 + 5 callers in the test file. 4) SESSION_SPEC.md #7: update the "Deadline = 3 × heartbeatInterval" bullet to reflect the new 1x semantics and default interval. Tests: - TestHandleSessionParameters_UpdatesIntervalAndDeadline: expect deadline ~= before+interval (1x) instead of before+3*interval. - TestHeartBeatLoop_HeartbeatsKeepInflightVRPCAlive: seed deadline at now+30ms (1x) instead of now+90ms (was 3x).

…Loop, heartBeatLoop) Third and final PR in the Session core stack. Adds the lifecycle orchestration that drives a Session from Start through teardown, plus the readLoop that dispatches server frames to the vRPC handlers landed in bigtable-session-vrpc. Stacks on bigtable-session-vrpc; deletes the minimal ForceClose stub that PR shipped, replacing it with the full lifecycle-shaped body. New file: session_lifecycle.go (~640 LOC) - Session.Start(ctx, OpenSessionRequest) — transitions New→Starting, Sends the OpenSession frame, fires onStart, spawns readLoop + heartBeatLoop. Wraps a failed Send as codes.Unavailable so retry plumbing treats pre-wire OpenSession loss the same as any other transport-side loss. - ForceClose — full body: transitionTo(Closed), setCloseReason, notifyClosing (once), cancelActiveRPCs, signalQuiescent, notifyClosed. - Close(ctx, CloseSessionRequest) — graceful drain: Ready→Closing, waits on quiescent, sends CloseSession, transitions to WaitServerClose, arms the pool's stuck-session sweep to eventually ForceClose if the server never confirms. - notifyClosing / notifyClosed — once-guarded hook dispatchers with the strict onClosing-precedes-onClose ordering enforced. - readLoop(ctx) — Recv-loop that dispatches every SessionResponse variant via handleSessionResponse, drives handleClose on stream termination, records msgsRecv counters + resets heartbeat deadline on every recognized frame. - handleSessionResponse — dispatch switch to handleOpenSession / handleVRPCResponse / handleErrorResponse / handleSessionParameters / handleGoAway / handleSessionRefreshConfig. Heartbeat frames update the deadline via resetHeartbeatDeadline; the unknown-payload branch records a debug tag but doesn't reset the deadline (so a misbehaving server can't keep the watchdog satisfied with junk). - handleOpenSession — Starting→Ready transition, PeerInfo extract from the bidi header (synchronous, matches Java's onHeaders synchrony), fires onActive. - handleGoAway / handleClose / handleErrorResponse / handleSessionParameters / handleSessionRefreshConfig — protocol-level handlers. - heartBeatLoop — Timer + heartbeatWake reactive-wake pair. Only enforces the deadline while a vRPC is in flight (idle sessions legitimately receive no server heartbeats). See SESSION_SPEC.md #7. - peerInfoExtracter — parses the bigtable-peer-info header, stamps s.peerInfo atomically before onActive fires. - closeReasonLabel / closeReasonToCause — CloseSessionRequest.Reason → string / sentinel error mapping for close-reason attribution. New file: session_lifecycle_test.go (~660 LOC) — 30+ tests covering Start, ForceClose, Close, readLoop, handleSessionResponse dispatch, handleOpenSession + peerInfo parsing, handleGoAway, handleClose, handleErrorResponse (rpc_id=0 harmlessly drops via routeVRPCFrame guards), heartBeatLoop (reactive wake, idle-gate, missed-heartbeat ForceClose). Edits to session_vrpc.go - Delete the minimal ForceClose stub introduced in the prior PR — the full body lives in session_lifecycle.go now. Edits to session_test.go - Add hookCounts helper (start/active/close callback counters). - Add setSlotForTest helper (seed the in-flight slot from tests). Stack: - googleapis#20211 — Session debug surface - googleapis#20213 — Session vRPC dispatch + slot lifecycle - This PR — Session lifecycle

sushanb added 3 commits January 7, 2026 00:16

feat(bigtable): use direct access if available

37b5ace

fix(bigtable): add feature flag

cd4eb3b

fix

2f53fa2

sushanb requested a review from nimf January 7, 2026 05:33

fix

69d1ad6

Navigation Menu

Repository navigation

feat(bigtable): Use direct access if it returns true. Guarded by CBT_BIGTABLE_CONN_POOL - #7

feat(bigtable): Use direct access if it returns true. Guarded by CBT_BIGTABLE_CONN_POOL#7
sushanb wants to merge 4 commits into
health_check_from
opt_conn

sushanb commented Jan 7, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

sushanb commented Jan 7, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant