Skip to content

fix: retry session/resume to absorb backend cold-start window #29

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
william0wang merged 1 commit into from
Jul 24, 2026
Merged

fix: retry session/resume to absorb backend cold-start window #29

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 37 additions & 6 deletions src/handlers/session.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,7 +130,6 @@ export async function resumeSession(
params: acp.ResumeSessionRequest,
cx: acp.AgentContext,
): Promise<acp.ResumeSessionResponse> {
const backend = server.ensureBackend();
const targetSid = params.sessionId;
const cwd = params.cwd ?? process.cwd();
if (!targetSid) throw new Error("sessionId required");
Expand All @@ -146,8 +145,7 @@ export async function resumeSession(
};
const runtimeModel = buildResumeRuntimeModel();
if (runtimeModel !== null) zcParams.runtimeModel = runtimeModel;
const resp = await backend.request(server.nextId(), "session/resume", zcParams, 15000);
if (resp.error) throw new Error(`zcode resume failed: ${resp.error.message ?? ""}`);
await resumeBackendSession(server, zcParams);

server.registerSession(targetSid, targetSid);
log(`session/resume -> ${targetSid}`);
Expand All @@ -172,7 +170,6 @@ export async function loadSession(
params: acp.LoadSessionRequest,
cx: acp.AgentContext,
): Promise<acp.LoadSessionResponse> {
const backend = server.ensureBackend();
const targetSid = params.sessionId;
const cwd = params.cwd ?? process.cwd();
if (!targetSid) throw new Error("sessionId required");
Expand All @@ -183,8 +180,7 @@ export async function loadSession(
};
const runtimeModel = buildResumeRuntimeModel();
if (runtimeModel !== null) zcParams.runtimeModel = runtimeModel;
const resp = await backend.request(server.nextId(), "session/resume", zcParams, 15000);
if (resp.error) throw new Error(`zcode resume failed: ${resp.error.message ?? ""}`);
await resumeBackendSession(server, zcParams);
server.registerSession(targetSid, targetSid);
log(`session/load → ${targetSid}`);
server.ensureBackgroundListener(targetSid);
Expand Down Expand Up @@ -712,6 +708,41 @@ function fileUriToPath(uri: string): string {
}
}

/**
* Resume a zcode session with retry on transient timeouts.
*
* The backend drops RPCs issued during its cold-start window (between process
* spawn and `startup.completed`). The first resume after a fresh backend spawn
* can land in that gap and time out without the backend ever seeing it. A single
* retry — issued after the startup window has elapsed — succeeds. Non-timeout
* errors (Invalid params, session not found) fail fast.
*/
async function resumeBackendSession(
server: ZcodeAcpServer,
zcParams: Record<string, unknown>,
): Promise<void> {
const backend = server.ensureBackend();
const MAX_ATTEMPTS = 2;
const ATTEMPT_TIMEOUT_MS = 15_000;
for (let attempt = 1; attempt <= MAX_ATTEMPTS; attempt++) {
const resp = await backend.request(
server.nextId(),
"session/resume",
zcParams,
ATTEMPT_TIMEOUT_MS,
);
if (!resp.error) return;
const isTimeout = resp.error.message === "timeout";
if (!isTimeout || attempt === MAX_ATTEMPTS) {
throw new Error(`zcode resume failed: ${resp.error.message ?? ""}`);
}
log(
`session/resume attempt ${attempt}/${MAX_ATTEMPTS} timed out, retrying (backend cold-start window)`,
);
await sleep(1000);
}
}

/** Fetch session/messages from zcode. */
async function fetchMessages(server: ZcodeAcpServer, zcodeSid: string): Promise<ZcodeMessage[]> {
const backend = server.ensureBackend();
Expand Down
Loading