fix(background-agent): add OMO_DISABLE_PROCESS_CLEANUP env opt-out for global handlers (fixes #3856)

Currently `registerManagerForCleanup` unconditionally installs global `uncaughtException` and `unhandledRejection` listeners that call `process.exit(1)` after cleanup. For users who load the plugin but never run background-agent tasks, these handlers turn transient streaming errors (e.g. undici `UND_ERR_SOCKET` mid-stream resets from `api.githubcopilot.com`) into a full process kill — opencode dies after every flaky response.

Add an `OMO_DISABLE_PROCESS_CLEANUP` env var (accepts 1/true/yes/on, case-insensitive) that skips just the error-event registration. Signal handlers (SIGINT/SIGTERM/SIGBREAK/beforeExit/exit) remain installed so graceful shutdown of any in-flight cleanup targets still runs. This is the lowest-risk near-term mitigation suggested in the issue (option #2): users opting in pay the cost of unhandled rejections themselves, but no longer lose their session to a transient socket reset.

Verification: 6 new test cases cover env-var precedence (set/unset, truthy/falsy values), signal-handler preservation, and behavior under `uncaughtException`. All 20 tests in process-cleanup.test.ts pass. Typecheck clean. Manual QA confirms env-var detection works end-to-end.
This commit is contained in:
MoerAI
2026-05-11 18:43:40 +09:00
parent 90f0971f4f
commit dd68f324d0
2 changed files with 126 additions and 0 deletions
@@ -235,6 +235,103 @@ describe("#given process cleanup registration", () => {
})
})
describe("#given OMO_DISABLE_PROCESS_CLEANUP env var", () => {
let originalEnvValue: string | undefined
beforeEach(() => {
originalEnvValue = process.env.OMO_DISABLE_PROCESS_CLEANUP
})
afterEach(() => {
if (originalEnvValue === undefined) {
delete process.env.OMO_DISABLE_PROCESS_CLEANUP
} else {
process.env.OMO_DISABLE_PROCESS_CLEANUP = originalEnvValue
}
})
test("#given env var is set to 1 #when registerManagerForCleanup runs #then uncaughtException handler is NOT registered", () => {
const uncaughtExceptionListenersBefore = process.listeners("uncaughtException")
const unhandledRejectionListenersBefore = process.listeners("unhandledRejection")
process.env.OMO_DISABLE_PROCESS_CLEANUP = "1"
const manager = { shutdown: mock(() => {}) }
registeredManagers.push(manager)
registerManagerForCleanup(manager)
expect(process.listeners("uncaughtException")).toHaveLength(uncaughtExceptionListenersBefore.length)
expect(process.listeners("unhandledRejection")).toHaveLength(unhandledRejectionListenersBefore.length)
})
test("#given env var is set to true #when registerManagerForCleanup runs #then handlers are NOT registered", () => {
const uncaughtExceptionListenersBefore = process.listeners("uncaughtException")
process.env.OMO_DISABLE_PROCESS_CLEANUP = "true"
const manager = { shutdown: mock(() => {}) }
registeredManagers.push(manager)
registerManagerForCleanup(manager)
expect(process.listeners("uncaughtException")).toHaveLength(uncaughtExceptionListenersBefore.length)
})
test("#given env var is set to 0 #when registerManagerForCleanup runs #then handlers ARE registered", () => {
const uncaughtExceptionListenersBefore = process.listeners("uncaughtException")
process.env.OMO_DISABLE_PROCESS_CLEANUP = "0"
const manager = { shutdown: mock(() => {}) }
registeredManagers.push(manager)
registerManagerForCleanup(manager)
expect(process.listeners("uncaughtException")).toHaveLength(uncaughtExceptionListenersBefore.length + 1)
})
test("#given env var is unset #when registerManagerForCleanup runs #then handlers ARE registered", () => {
const uncaughtExceptionListenersBefore = process.listeners("uncaughtException")
delete process.env.OMO_DISABLE_PROCESS_CLEANUP
const manager = { shutdown: mock(() => {}) }
registeredManagers.push(manager)
registerManagerForCleanup(manager)
expect(process.listeners("uncaughtException")).toHaveLength(uncaughtExceptionListenersBefore.length + 1)
})
test("#given env var is set #when signals fire #then SIGINT/SIGTERM/beforeExit/exit handlers still run cleanup", () => {
const exitListenersBefore = process.listeners("exit")
process.env.OMO_DISABLE_PROCESS_CLEANUP = "yes"
const shutdown = mock(() => {})
const manager = { shutdown }
registeredManagers.push(manager)
registerManagerForCleanup(manager)
const exitListener = getNewListener("exit", exitListenersBefore)
exitListener()
expect(shutdown).toHaveBeenCalledTimes(1)
})
test("#given env var is set AND process emits uncaughtException #when event fires #then manager shutdown is NOT invoked by our handler", async () => {
process.env.OMO_DISABLE_PROCESS_CLEANUP = "1"
const exitSpy = spyOn(process, "exit").mockImplementation((() => undefined) as never)
const shutdown = mock(() => {})
const manager = { shutdown }
registeredManagers.push(manager)
try {
registerManagerForCleanup(manager)
// Other listeners on uncaughtException may exist (e.g. node default).
// We assert that OUR handler did not run cleanup.
process.emit("uncaughtException", new Error("boom"))
await flushMicrotasks()
expect(shutdown).not.toHaveBeenCalled()
} finally {
exitSpy.mockRestore()
}
})
})
describe("#given uncaught exception and rejection cleanup", () => {
test("#given manager registered AND process emits uncaughtException #when event fires #then manager shuts down before process exits", async () => {
const exitSpy = spyOn(process, "exit").mockImplementation((() => undefined) as never)
@@ -3,6 +3,26 @@ import { log } from "../../shared"
type ProcessCleanupSignal = NodeJS.Signals | "beforeExit" | "exit"
type ProcessCleanupErrorEvent = "uncaughtException" | "unhandledRejection"
/**
* When set to a truthy value (1/true/yes/on), suppresses the global
* uncaughtException / unhandledRejection handlers that force-exit the host
* process. Use this when the plugin is installed but background-agent tasks
* are not actively in use, to avoid OpenCode dying on transient streaming
* errors propagated as unhandled rejections (see issue #3856).
*
* Signal handlers (SIGINT/SIGTERM/SIGBREAK/beforeExit/exit) remain registered
* because they are needed for graceful shutdown of any in-flight cleanup
* targets that were registered before the user noticed the issue.
*/
const PROCESS_CLEANUP_DISABLE_ENV = "OMO_DISABLE_PROCESS_CLEANUP"
const TRUTHY_ENV_VALUES = new Set(["1", "true", "yes", "on"])
function isProcessCleanupErrorHandlersDisabled(): boolean {
const raw = process.env[PROCESS_CLEANUP_DISABLE_ENV]
if (!raw) return false
return TRUTHY_ENV_VALUES.has(raw.trim().toLowerCase())
}
/** @internal test-only seam: prevents process.exitCode from contaminating bun test runner */
let _scheduleForcedExitEnabled = true
@@ -116,6 +136,15 @@ export function registerManagerForCleanup(manager: CleanupTarget): void {
}
registerSignal("beforeExit", false)
registerSignal("exit", false)
if (isProcessCleanupErrorHandlersDisabled()) {
log(
`[background-agent] ${PROCESS_CLEANUP_DISABLE_ENV} is set; skipping global uncaughtException/unhandledRejection handler registration. `
+ "Signal handlers (SIGINT/SIGTERM/beforeExit/exit) remain active.",
)
return
}
cleanupErrorHandlers.set("uncaughtException", registerErrorEvent("uncaughtException", cleanupAll))
cleanupErrorHandlers.set("unhandledRejection", registerErrorEvent("unhandledRejection", cleanupAll))
}