Make telemetry reliable across interrupts and headless installs

Three reliability gaps surfaced while auditing why PostHog numbers were
untrustworthy:

1. Interrupted commands lost their events. capture() is fire-and-forget and the
   only flush guarantee lived in a finally block, which SIGINT/SIGTERM skip — so
   Ctrl-C'ing a long ingest or an MCP client killing 'ktx mcp stdio' dropped the
   command event and any queued events. Add SIGINT/SIGTERM handlers (real-process
   entry only; never under test/programmatic io) that mark the active command
   span aborted, emit it, drain the emitter, then exit. Idempotent with the
   normal finally path via the single-consume command span.

2. Headless-first installs were invisible. loadTelemetryIdentity refused to mint
   an installId unless stdout was a TTY, so a machine whose first run was an
   IDE-launched MCP server or a script emitted nothing, ever. Mint on first run
   regardless of surface (still honoring CI/DO_NOT_TRACK/KTX_TELEMETRY_DISABLED),
   writing the one-time notice to stderr — safe under the MCP stdio protocol,
   which reserves stdout. Drop the now-unused stdoutIsTTY option.

3. No guard against silent emit regressions (the 0.7.0 scan_completed blackout).
   Add tests: the shared executePublicIngestTarget chokepoint emits exactly one
   ingest_completed on success and on the preflight-failure branch, and a
   database target invokes the scan that emits scan_completed; plus coverage for
   the aborted-flush helper.

Identity is unchanged otherwise: every event still attributes to the installId
in ~/.ktx/telemetry.json. No event/field changes, so Node<->Python schema parity
is untouched. Docs updated to reflect first-run-on-any-surface activation.
This commit is contained in:
Andrey Avtomonov 2026-06-02 23:19:37 +02:00
parent 2334a4b6e3
commit cb6a67c2d7
7 changed files with 219 additions and 66 deletions

View file

@ -89,6 +89,46 @@ export async function runInitForCommander(
return await runInit(args, io);
}
function signalExitCode(signal: NodeJS.Signals): number {
// 128 + signal number: SIGINT (2) -> 130, SIGTERM (15) -> 143.
return signal === 'SIGTERM' ? 143 : 130;
}
/**
* Flush telemetry on interrupt for the real CLI process. `capture()` is
* fire-and-forget and the only flush guarantee lives in a `finally` a signal
* skips, so Ctrl-C / `kill` of a long-running command (ingest, `mcp stdio`)
* would otherwise drop its `command` event and queued events. Installed only
* when driving the actual process; programmatic/test callers pass their own
* `io` and never reach here. Returns a disposer that removes the listeners.
*/
function installTelemetrySignalFlush(io: KtxCliIo, info: KtxCliPackageInfo): () => void {
let handling = false;
const handle = (signal: NodeJS.Signals): void => {
if (handling) {
process.exit(signalExitCode(signal));
}
handling = true;
void (async () => {
try {
const { emitAbortedCommandAndShutdown } = await import('./telemetry/index.js');
await emitAbortedCommandAndShutdown({ packageInfo: info, io });
} catch {
// Best-effort: never let a telemetry hiccup block the interrupt exit.
}
process.exit(signalExitCode(signal));
})();
};
const onSigint = (): void => handle('SIGINT');
const onSigterm = (): void => handle('SIGTERM');
process.on('SIGINT', onSigint);
process.on('SIGTERM', onSigterm);
return () => {
process.off('SIGINT', onSigint);
process.off('SIGTERM', onSigterm);
};
}
export async function runKtxCli(
argv = process.argv.slice(2),
io: KtxCliIo = process,
@ -98,7 +138,14 @@ export async function runKtxCli(
profileMark('runtime:runKtxCli');
const { runCommanderKtxCli } = await profileSpan('import ./cli-program.js', () => import('./cli-program.js'));
return await runCommanderKtxCli(argv, io, deps, info, {
runInit: runInitForCommander,
});
// Real-process entry only: flush telemetry if interrupted. Test/programmatic
// callers pass their own `io`, so they never install process-level handlers.
const removeSignalFlush = (io as unknown) === process ? installTelemetrySignalFlush(io, info) : undefined;
try {
return await runCommanderKtxCli(argv, io, deps, info, {
runInit: runInitForCommander,
});
} finally {
removeSignalFlush?.();
}
}