Coder/deepseek-harness
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(e2b): harden remote teardown transactions

Browse Source
This commit is contained in:
Tianyi Cui
2026-07-29 07:03:04 +08:00
parent 62df6d0c09
commit ec0310ca1b
19 changed files with 712 additions and 135 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
packages/e2b/subprocess-e2b/README.i18n.yaml
View File

@@ -2,5 +2,5 @@
# side as of the last confirmed-consistent state. Both languages carry equal authority;
# after editing either side, bring the other along and re-record with:
# pnpm run verify-translation-pairing --write packages/e2b/subprocess-e2b/README.md
README.md: 58f13b264324d2168d4bc264b60100cdafa0a366
README.zh.md: d547e4e8d66f629edcef753651180fe1806ad9bf
README.md: b1aa480305fef9ca485340b6bdae50f9fcffd88f
README.zh.md: b4824c90e7fb908d9c12e77c19d2b952055dcb40

12
packages/e2b/subprocess-e2b/README.md
View File

@@ -6,14 +6,14 @@ E2B implementation of the [`@deepseek-ai/dsh-subprocess`](../../subprocess/subpr
## Behavior
- **Asynchronous remote start** — the synchronous seam returns a handle immediately while `Sandbox.commands.run(..., { background: true })` starts remotely. `pid` is `-1` until the wrapper publishes and the adapter validates its process-group id; `done`, stdin, termination, and `waitForExit()` wait for readiness internally.
- **Asynchronous remote start** — the synchronous seam returns a handle immediately while `Sandbox.commands.run(..., { background: true })` starts remotely. `pid` is `-1` until the wrapper publishes and the adapter validates its process-group id; stdin and ordinary observation wait for that publication, while cancellation can stop the provisional SDK handle first.
- **Execution-world coordinates** — `cwd` and private `runtimeRoot` come from the shared owner; executable lookup verifies absolute paths or resolves a bare name against the sandbox PATH plus explicit overrides.
- **Linux process groups** — a quoted wrapper starts each argv under `exec setsid --wait` and records its actual process-group id plus private status files beneath `ctx.e2b.runtimeRoot/processes`. The handle waits for that file instead of treating the SDK command PID as its published identity. Termination signals the negative recorded id with `SIGTERM`, waits the caller's `graceMs`, then escalates to `SIGKILL` and the SDK kill fallback. If publication fails, the SDK PID remains the provisional `exec setsid` group id; rollback kills and verifies that group before startup rejects. Service disposal terminates and joins every retained handle before the sandbox owner disposes.
- **Environment boundary** — the wrapper starts from the sandbox command environment, removes ambient `DSH_*` and credential-shaped (`*KEY*`, `*SECRET*`, `*TOKEN*`) names, then restores every `spec.env` entry as an explicit caller opt-in. Host ambient variables never enter the sandbox implicitly.
- **Stdio projection** — the remote wrapper branches raw bytes into optional bounded spill files, frames each live chunk as newline-delimited base64 ASCII, and the host incrementally restores bytes across arbitrary SDK callback boundaries. Pipe mode writes those bytes to host Node streams; inherit mode writes them to the harness process streams; collect mode retains a bounded host tail with offset reads. Batch and streaming stdin use the SDK handle.
- **Terminal sessions** — `spawnTerminal()` uses E2B's byte PTY API, installs the exact argv and scrubbed environment through private mode-`0600` files, reports the foreground process group, sends real signals, and tears down every group in the remote terminal session before settlement. Prompt detection, scrollback, readiness, and owner policy remain in `dsh-pty-local`.
- **Linux process groups** — a quoted wrapper starts each argv under `exec setsid --wait` and records its actual process-group id plus private status files beneath `ctx.e2b.runtimeRoot/processes`. The handle waits for that file instead of treating the SDK command PID as its published identity. Termination signals the negative recorded id with `SIGTERM`, waits the caller's `graceMs`, then escalates to `SIGKILL` and the SDK kill fallback; TERM delivery or probe failures also force that escalation. A failed transaction is observable through `waitForExit()` and may be retried. Before publication, cancellation uses the provisional SDK handle; if publication fails, rollback kills and verifies the provisional group before startup rejects. Service disposal terminates and joins every retained handle before the sandbox owner disposes.
- **Environment boundary** — the wrapper starts from the sandbox command environment, removes ambient `DSH_*` and credential-shaped (`*KEY*`, `*SECRET*`, `*TOKEN*`) names, then restores every `spec.env` entry as an explicit caller opt-in. Host ambient variables never enter the sandbox implicitly. Private environment files are removed after consumption, and failed command or terminal setup removes its private state before rejecting.
- **Stdio projection** — the remote wrapper branches raw bytes into optional bounded spill files, frames each live chunk as newline-delimited base64 ASCII, and the host incrementally restores bytes across arbitrary SDK callback boundaries. Pipe mode writes those bytes to host Node streams; inherit mode writes them to the harness process streams; collect mode retains a bounded host tail with offset reads. The wrapper publishes the direct command status before waiting for inherited writers; after `graceMs`, the adapter disconnects an incomplete SDK stream, withholds its partial spill, and returns that status while retaining the remote group for `waitForExit()` and termination. Batch and streaming stdin use the SDK handle.
- **Terminal sessions** — `spawnTerminal()` uses E2B's byte PTY API, installs the exact argv and scrubbed environment through private mode-`0600` files, reports the foreground process group, sends real signals, and tears down every group in the remote terminal session before settlement. Setup and teardown own the private state transaction, including failure cleanup. Prompt detection, scrollback, readiness, and owner policy remain in `dsh-pty-local`.
The base E2B image supplies the runtime and Bash/GNU utilities this adapter invokes: `node`, `bash`, `setsid`, `ps`, `awk`, `tr`, `env`, `chmod`, `tee`, `head`, and `kill`. A custom template must retain compatible commands and E2B PTY support.
The base E2B image supplies the runtime and Bash/GNU utilities this adapter invokes: `node`, `bash`, `setsid`, `ps`, `awk`, `tr`, `env`, `chmod`, `tee`, `head`, `rm`, and `kill`. A custom template must retain compatible commands and E2B PTY support.
## Model Experience

12
packages/e2b/subprocess-e2b/README.zh.md
View File

@@ -6,14 +6,14 @@
## 行为
- **异步远程启动**：同步 seam 会立即返回一个句柄，同时由 `Sandbox.commands.run(..., { background: true })` 在远程启动进程。包装层发布进程组 ID 并由适配器完成验证之前，`pid` 为 `-1`；`done`、stdin、终止和 `waitForExit()` 会在内部等待就绪。
- **异步远程启动**：同步 seam 会立即返回一个句柄，同时由 `Sandbox.commands.run(..., { background: true })` 在远程启动进程。包装层发布进程组 ID 并由适配器完成验证之前，`pid` 为 `-1`；stdin 和常规观察会等待该发布，而取消操作可以先停止临时 SDK 句柄。
- **执行世界坐标**：`cwd` 和私有 `runtimeRoot` 来自共享所有者；可执行文件查找会验证绝对路径，或根据沙箱 PATH 加显式覆盖来解析裸名称。
- **Linux 进程组**：带引号保护的包装层会在 `exec setsid --wait` 下启动每组 argv，并在 `ctx.e2b.runtimeRoot/processes` 下记录实际进程组 ID 和私有状态文件。句柄会等待该文件，而不会把 SDK 命令 PID 当作已发布的身份。终止操作以记录的负数 ID 发送 `SIGTERM`，等待调用方的 `graceMs`，再升级到 `SIGKILL` 和 SDK kill 回退。如果发布失败，SDK PID 仍为临时的 `exec setsid` 进程组 ID；回滚会终止并验证该进程组，随后启动操作才会以拒绝结束。服务 dispose（资源释放）会在沙箱所有者释放前终止并等待每个保留句柄退出。
- **环境边界**：包装层从沙箱命令环境开始，移除环境中的 `DSH_*` 和形似凭据的名称（`*KEY*`、`*SECRET*`、`*TOKEN*`），再把每个 `spec.env` 条目恢复为调用方显式选择。宿主环境变量绝不会隐式进入沙箱。
- **stdio 投影**：远程包装层先把原始字节分流到可选的有界 spill 文件，再把每个实时分片编码为换行分隔的 base64 ASCII 帧；宿主会跨任意 SDK 回调边界增量恢复字节。pipe 模式把这些字节写入宿主 Node 流；inherit 模式把字节写入 harness 进程流；collect 模式保留有界的宿主尾部，并支持基于偏移量读取。批量 stdin 和流式 stdin 都使用 SDK 句柄。
- **终端会话**：`spawnTerminal()` 使用 E2B 的字节 PTY API，以 mode 为 `0600` 的私有文件传入原样 argv 与清理后的环境，报告前台进程组，发送真实信号，并在结算前清理远程终端会话中的每个进程组。提示符检测、scrollback、就绪状态与所有者策略仍归 `dsh-pty-local` 所有。
- **Linux 进程组**：带引号保护的包装层会在 `exec setsid --wait` 下启动每组 argv，并在 `ctx.e2b.runtimeRoot/processes` 下记录实际进程组 ID 和私有状态文件。句柄会等待该文件，而不会把 SDK 命令 PID 当作已发布的身份。终止操作以记录的负数 ID 发送 `SIGTERM`，等待调用方的 `graceMs`，再升级到 `SIGKILL` 和 SDK kill 回退；TERM 信号发送或探测失败也会强制触发该升级。失败的事务可通过 `waitForExit()` 观察，并可重试。发布前，取消操作使用临时 SDK 句柄；如果发布失败，回滚会终止并验证临时进程组，随后启动操作才会拒绝。服务 dispose（资源释放）会在沙箱所有者释放前终止并等待每个保留句柄退出。
- **环境边界**：包装层从沙箱命令环境开始，移除环境中的 `DSH_*` 和形似凭据的名称（`*KEY*`、`*SECRET*`、`*TOKEN*`），再把每个 `spec.env` 条目恢复为调用方显式选择。宿主环境变量绝不会隐式进入沙箱。私有环境文件在使用后会被删除；命令或终端设置失败时，会先删除其私有状态再拒绝。
- **stdio 投影**：远程包装层先把原始字节分流到可选的有界 spill 文件，再把每个实时分片编码为换行分隔的 base64 ASCII 帧；宿主会跨任意 SDK 回调边界增量恢复字节。pipe 模式把这些字节写入宿主 Node 流；inherit 模式把字节写入 harness 进程流；collect 模式保留有界的宿主尾部，并支持基于偏移量读取。包装层会在等待继承管道的写入方之前发布直接命令状态；超过 `graceMs` 后，适配器会断开未完成的 SDK 流，不公开其中不完整的 spill，并返回该状态，同时保留远程进程组供 `waitForExit()` 和终止操作使用。批量 stdin 和流式 stdin 都使用 SDK 句柄。
- **终端会话**：`spawnTerminal()` 使用 E2B 的字节 PTY API，以 mode 为 `0600` 的私有文件传入原样 argv 与清理后的环境，报告前台进程组，发送真实信号，并在结算前清理远程终端会话中的每个进程组。setup 与 teardown 负责私有状态事务，包括失败清理。提示符检测、scrollback、就绪状态与所有者策略仍归 `dsh-pty-local` 所有。
基础 E2B 镜像提供该适配器调用的运行时和 Bash/GNU 工具：`node`、`bash`、`setsid`、`ps`、`awk`、`tr`、`env`、`chmod`、`tee`、`head` 和 `kill`。自定义模板必须保留兼容的命令和 E2B PTY 支持。
基础 E2B 镜像提供该适配器调用的运行时和 Bash/GNU 工具：`node`、`bash`、`setsid`、`ps`、`awk`、`tr`、`env`、`chmod`、`tee`、`head`、`rm` 和 `kill`。自定义模板必须保留兼容的命令和 E2B PTY 支持。
## 模型体验

6
packages/e2b/subprocess-e2b/src/index.ts
View File

@@ -46,14 +46,12 @@ export class E2BSubprocessService extends SubprocessService {
const pending: Promise<unknown>[] = []
for (const handle of handles) {
handle.terminate()
pending.push(handle.done.catch(() => {}).then(() => handle.waitForExit()))
pending.push(handle.waitForExit().then(() => { this.live.delete(handle) }))
}
for (const terminal of terminals) {
terminal.terminate()
pending.push(terminal.waitForExit())
pending.push(terminal.waitForExit().then(() => { this.terminals.delete(terminal) }))
}
this.live.clear()
this.terminals.clear()
await Promise.all(pending)
}, 'e2b subprocess teardown')
}

8
packages/e2b/subprocess-e2b/src/output.ts
View File

@@ -66,6 +66,7 @@ export class E2BOutputReader implements SubprocessOutputReader {
private chunks: Buffer[] = []
private retainedBytes = 0
private totalBytes = 0
private spillValid = true
/**
* Create a bounded reader over one remote spill path.
@@ -84,6 +85,11 @@ export class E2BOutputReader implements SubprocessOutputReader {
return this.totalBytes
}
/** Stop advertising a remote spill whose writer did not reach clean EOF. */
invalidateSpill(): void {
this.spillValid = false
}
/**
* Append one byte-faithful decoded transport event.
* @param bytes - Raw command bytes recovered from the ASCII SDK transport.
@@ -120,7 +126,7 @@ export class E2BOutputReader implements SubprocessOutputReader {
text: retained.subarray(start).toString('utf8'),
nextOffset: this.totalBytes,
lossy,
...(lossy && this.maxSpillBytes !== undefined && this.totalBytes <= this.maxSpillBytes
...(lossy && this.spillValid && this.maxSpillBytes !== undefined && this.totalBytes <= this.maxSpillBytes
? { spillPath: this.spillPath }
: {}),
}

332
packages/e2b/subprocess-e2b/src/process.ts
View File

@@ -5,6 +5,7 @@ import { PassThrough, Writable } from 'node:stream'
import { posix } from 'node:path'
import {
CommandExitError,
FileNotFoundError,
quoteE2BShellArg,
} from '@deepseek-ai/dsh-e2b'
import type { CommandHandle, CommandResult, Sandbox } from '@deepseek-ai/dsh-e2b'
@@ -41,6 +42,10 @@ function hasSpill(mode: SubprocessOutputMode): mode is SubprocessCollect & { spi
return isCollect(mode) && mode.spill !== undefined
}
function isValidProcessId(value: number): boolean {
return Number.isSafeInteger(value) && value > 0
}
function asError(error: unknown): Error {
return error instanceof Error ? error : new Error(String(error))
}
@@ -73,6 +78,20 @@ interface RemotePaths {
stderr: string
}
type CommandSettlement =
| { kind: 'result'; result: CommandResult }
| { kind: 'error'; error: unknown }
function withinMs<T>(promise: Promise<T>, timeoutMs: number): Promise<T | undefined> {
return new Promise<T | undefined>((resolve) => {
const timer = setTimeout(() => { resolve(undefined) }, timeoutMs)
void promise.then((value) => {
clearTimeout(timer)
resolve(value)
})
})
}
function remoteEnvironment(raw: string, explicit: Readonly<Record<string, string>> | undefined): string {
const environment = scrubRemoteEnvironment(raw)
for (const [name, value] of Object.entries(explicit ?? {})) environment.set(name, value)
@@ -96,15 +115,16 @@ function commandText(spec: SubprocessSpawnSpec, paths: RemotePaths): string {
'dsh_e2b_tr=$4',
'dsh_e2b_tee=$5',
'dsh_e2b_head=$6',
'shift 6',
'dsh_e2b_rm=$7',
'shift 7',
'dsh_e2b_pgid="$("$dsh_e2b_ps" -o pgid= -p "$$" | "$dsh_e2b_tr" -d " ")"',
`printf '%s\\n' "$dsh_e2b_pgid" > ${quoteE2BShellArg(paths.pid)}`,
`mapfile -d '' -t dsh_e2b_env < ${quoteE2BShellArg(paths.environment)}`,
`: > ${quoteE2BShellArg(paths.environment)}`,
`"$dsh_e2b_rm" -f -- ${quoteE2BShellArg(paths.environment)}`,
`"$dsh_e2b_env_bin" -i "\${dsh_e2b_env[@]}" "$@" ${stdoutRedirect} ${stderrRedirect}`.trimEnd(),
'dsh_e2b_status=$?',
'wait',
`printf '%s\\n' "$dsh_e2b_status" > ${quoteE2BShellArg(paths.status)}`,
'wait',
'exit "$dsh_e2b_status"',
].join('\n')
const argv = spec.argv.map(quoteE2BShellArg).join(' ')
@@ -118,10 +138,11 @@ function commandText(spec: SubprocessSpawnSpec, paths: RemotePaths): string {
'dsh_e2b_tr="$(command -v tr)"',
'dsh_e2b_tee="$(command -v tee)"',
'dsh_e2b_head="$(command -v head)"',
'for dsh_e2b_tool in "$dsh_e2b_env_bin" "$dsh_e2b_setsid" "$dsh_e2b_bash" "$dsh_e2b_node" "$dsh_e2b_ps" "$dsh_e2b_tr" "$dsh_e2b_tee" "$dsh_e2b_head"; do',
'dsh_e2b_rm="$(command -v rm)"',
'for dsh_e2b_tool in "$dsh_e2b_env_bin" "$dsh_e2b_setsid" "$dsh_e2b_bash" "$dsh_e2b_node" "$dsh_e2b_ps" "$dsh_e2b_tr" "$dsh_e2b_tee" "$dsh_e2b_head" "$dsh_e2b_rm"; do',
' [[ "$dsh_e2b_tool" == /* && -x "$dsh_e2b_tool" ]] || exit 125',
'done',
`exec "$dsh_e2b_env_bin" -i "\${dsh_e2b_env[@]}" "$dsh_e2b_setsid" --wait -- "$dsh_e2b_bash" -c ${quoteE2BShellArg(inner)} dsh-e2b "$dsh_e2b_env_bin" "$dsh_e2b_node" "$dsh_e2b_ps" "$dsh_e2b_tr" "$dsh_e2b_tee" "$dsh_e2b_head" ${argv}`,
`exec "$dsh_e2b_env_bin" -i "\${dsh_e2b_env[@]}" "$dsh_e2b_setsid" --wait -- "$dsh_e2b_bash" -c ${quoteE2BShellArg(inner)} dsh-e2b "$dsh_e2b_env_bin" "$dsh_e2b_node" "$dsh_e2b_ps" "$dsh_e2b_tr" "$dsh_e2b_tee" "$dsh_e2b_head" "$dsh_e2b_rm" ${argv}`,
].join('\n')
return bootstrap
}
@@ -149,6 +170,26 @@ function waitTick(signal?: AbortSignal): Promise<boolean> {
})
}
const WAIT_ABORTED = Symbol('wait aborted')
function waitWithSignal<T>(promise: Promise<T>, signal: AbortSignal | undefined): Promise<T | typeof WAIT_ABORTED> {
if (signal === undefined) return promise
if (signal.aborted) return Promise.resolve(WAIT_ABORTED)
return new Promise<T | typeof WAIT_ABORTED>((resolve, reject) => {
const onAbort = (): void => { cleanup(); resolve(WAIT_ABORTED) }
const cleanup = (): void => { signal.removeEventListener('abort', onAbort) }
signal.addEventListener('abort', onAbort, { once: true })
if (signal.aborted) {
onAbort()
return
}
void promise.then(
(value) => { cleanup(); resolve(value) },
(error: unknown) => { cleanup(); reject(asError(error)) },
)
})
}
/** E2B-backed subprocess handle with deferred remote PID acquisition. */
export class E2BSubprocessHandle implements SubprocessHandle {
readonly stdin: Writable | undefined
@@ -157,6 +198,7 @@ export class E2BSubprocessHandle implements SubprocessHandle {
readonly collected: SubprocessHandle['collected']
readonly done: Promise<SubprocessOutcome>
private readonly commandState = Promise.withResolvers<CommandHandle | undefined>()
private readonly readyState = Promise.withResolvers<CommandHandle>()
private readonly stdoutDecoder = new E2BBase64Decoder()
private readonly stderrDecoder = new E2BBase64Decoder()
@@ -166,9 +208,14 @@ export class E2BSubprocessHandle implements SubprocessHandle {
private remotePid = -1
private commandHandle: CommandHandle | undefined
private outputTransportError: Error | undefined
private terminationRequested = false
private outputDrainExpired = false
private stateDirectoryCreated = false
private invalidHandleQuiescent = false
private provisionalHandleQuiescent = false
private terminationStarted = false
private terminationAttempt: Promise<void> | undefined
private terminationFailure: Error | undefined
private terminationSignal: NodeJS.Signals | null = null
private termination: Promise<void> | undefined
/**
* Begin an E2B command without blocking the synchronous subprocess spawn seam.
@@ -217,22 +264,47 @@ export class E2BSubprocessHandle implements SubprocessHandle {
/** @inheritdoc */
terminate(): void {
if (this.terminationRequested) return
this.terminationRequested = true
this.termination = this.terminateRemote()
void this.termination.catch(() => {})
if (this.terminationAttempt !== undefined) return
this.terminationStarted = true
this.terminationFailure = undefined
const attempt = this.terminateRemote()
this.terminationAttempt = attempt
void attempt.then(
() => {
this.terminationAttempt = undefined
},
(error: unknown) => {
this.terminationFailure = asError(error)
this.terminationAttempt = undefined
},
)
}
/** @inheritdoc */
async waitForExit(signal?: AbortSignal): Promise<boolean> {
let handle: CommandHandle | undefined
try {
handle = await this.readyForWait(signal)
} catch {
handle = this.commandHandle
if (this.terminationStarted) {
const observed = await waitWithSignal(this.commandState.promise, signal)
if (observed === WAIT_ABORTED) return false
handle = observed
if (handle === undefined) return true
if (this.remotePid <= 0) {
const attempt = this.terminationAttempt
if (attempt !== undefined && await waitWithSignal(attempt, signal) === WAIT_ABORTED) return false
this.throwTerminationFailure()
if (this.invalidHandleQuiescent || this.provisionalHandleQuiescent) return true
}
} else {
try {
const observed = await waitWithSignal(this.readyState.promise, signal)
if (observed === WAIT_ABORTED) return false
handle = observed
} catch {
handle = this.commandHandle
if (handle === undefined) return true
}
}
if (handle === undefined) return false
this.throwTerminationFailure()
let sandbox: Sandbox
try {
sandbox = await this.runtime.getSandbox()
@@ -242,33 +314,19 @@ export class E2BSubprocessHandle implements SubprocessHandle {
}
const processGroupId = this.remotePid > 0 ? this.remotePid : handle.pid
while (await this.groupAlive(sandbox, processGroupId, signal)) {
this.throwTerminationFailure()
if (!await waitTick(signal)) return false
}
this.throwTerminationFailure()
return !isAborted(signal)
}
private readyForWait(signal: AbortSignal | undefined): Promise<CommandHandle | undefined> {
if (signal === undefined) return this.readyState.promise
return new Promise<CommandHandle | undefined>((resolve, reject) => {
const onAbort = (): void => { cleanup(); resolve(undefined) }
const cleanup = (): void => { signal.removeEventListener('abort', onAbort) }
signal.addEventListener('abort', onAbort, { once: true })
if (signal.aborted) {
onAbort()
return
}
void this.readyState.promise.then(
(handle) => { cleanup(); resolve(handle) },
(error: unknown) => { cleanup(); reject(asError(error)) },
)
})
}
private readonly onAbort = (): void => { this.terminate() }
private async run(): Promise<SubprocessOutcome> {
let sandbox: Sandbox | undefined
try {
const sandbox = await this.runtime.getSandbox()
sandbox = await this.runtime.getSandbox()
await this.prepareState(sandbox)
const handle = await sandbox.commands.run(
commandText(this.spec, this.paths),
@@ -281,12 +339,26 @@ export class E2BSubprocessHandle implements SubprocessHandle {
onStderr: async (data) => { await this.dispatchOutput('stderr', data) },
},
)
if (!Number.isSafeInteger(handle.pid) || handle.pid <= 0) {
throw new Error(`subprocess-e2b: E2B returned invalid command pid ${handle.pid}`)
}
this.commandHandle = handle
const completion = handle.wait()
void completion.catch(() => {})
if (!isValidProcessId(handle.pid)) {
const invalidPid = new Error(`subprocess-e2b: E2B returned invalid command pid ${handle.pid}`)
try {
await handle.kill()
this.invalidHandleQuiescent = true
this.commandHandle = undefined
} catch (cleanupError: unknown) {
this.terminationFailure = asError(cleanupError)
this.commandState.resolve(handle)
throw new AggregateError(
[invalidPid, cleanupError],
'subprocess-e2b: invalid command pid rollback did not reach quiescence',
)
}
throw invalidPid
}
this.commandState.resolve(handle)
try {
this.remotePid = await this.waitForProcessGroupId(sandbox, completion)
} catch (error: unknown) {
@@ -302,16 +374,28 @@ export class E2BSubprocessHandle implements SubprocessHandle {
}
this.readyState.resolve(handle)
await this.writeBatchStdin(handle)
const outcome = await this.waitForCommand(completion)
const outcome = await this.waitForCommand(sandbox, handle, completion)
if (this.outputTransportError !== undefined) throw this.outputTransportError
const requireCompleteOutput = this.terminationSignal === null
const requireCompleteOutput = this.terminationSignal === null && !this.outputDrainExpired
this.stdoutDecoder.finish(requireCompleteOutput)
this.stderrDecoder.finish(requireCompleteOutput)
await this.finalizeSpills(sandbox)
return outcome
} catch (error: unknown) {
this.readyState.reject(error)
throw error
this.commandState.resolve(undefined)
let failure = error
if (sandbox !== undefined && this.stateDirectoryCreated) {
try {
await this.removeFailedState(sandbox)
} catch (cleanupError: unknown) {
failure = new AggregateError(
[error, cleanupError],
'subprocess-e2b: command failed and private state cleanup failed',
)
}
}
this.readyState.reject(failure)
throw failure
} finally {
this.spec.signal?.removeEventListener('abort', this.onAbort)
this.stdout?.end()
@@ -322,6 +406,7 @@ export class E2BSubprocessHandle implements SubprocessHandle {
private async prepareState(sandbox: Sandbox): Promise<void> {
const ambient = await sandbox.commands.run('env -0')
await sandbox.files.makeDir(this.stateDir)
this.stateDirectoryCreated = true
await sandbox.commands.run(`chmod 700 -- ${quoteE2BShellArg(this.stateDir)}`)
const files = [
{ path: this.paths.pid, data: '' },
@@ -405,20 +490,45 @@ export class E2BSubprocessHandle implements SubprocessHandle {
}
}
private async waitForCommand(completion: Promise<CommandResult>): Promise<SubprocessOutcome> {
try {
const result = await completion
return { exitCode: result.exitCode, signal: null }
} catch (error: unknown) {
if (error instanceof CommandExitError) {
return this.terminationSignal === null
? { exitCode: error.exitCode, signal: null }
: { exitCode: null, signal: this.terminationSignal }
private async waitForCommand(
sandbox: Sandbox,
handle: CommandHandle,
completion: Promise<CommandResult>,
): Promise<SubprocessOutcome> {
const settlement = completion.then<CommandSettlement, CommandSettlement>(
result => ({ kind: 'result', result }),
(error: unknown) => ({ kind: 'error', error }),
)
while (true) {
const rawStatus = (await sandbox.files.read(this.paths.status)).trim()
if (rawStatus.length > 0) {
const exitCode = Number(rawStatus)
if (!/^(?:0|[1-9][0-9]*)$/.test(rawStatus) || !Number.isSafeInteger(exitCode) || exitCode > 255) {
throw new Error(`subprocess-e2b: remote wrapper published invalid exit code ${JSON.stringify(rawStatus)}`)
}
const completed = await withinMs(settlement, this.spec.graceMs)
if (completed !== undefined) return this.commandOutcome(completed)
this.outputDrainExpired = true
this.stdoutReader?.invalidateSpill()
this.stderrReader?.invalidateSpill()
await handle.disconnect()
return { exitCode, signal: null }
}
throw error
const completed = await Promise.race([settlement, waitTick().then(() => undefined)])
if (completed !== undefined) return this.commandOutcome(completed)
}
}
private commandOutcome(settlement: CommandSettlement): SubprocessOutcome {
if (settlement.kind === 'result') return { exitCode: settlement.result.exitCode, signal: null }
if (settlement.error instanceof CommandExitError) {
return this.terminationSignal === null
? { exitCode: settlement.error.exitCode, signal: null }
: { exitCode: null, signal: this.terminationSignal }
}
throw settlement.error
}
private async rollbackUnpublishedGroup(sandbox: Sandbox, handle: CommandHandle): Promise<void> {
// The bootstrap ends in an exec chain through the scrubbed environment and
// `setsid`, so E2B's command PID is the provisional group id even before the
@@ -433,35 +543,99 @@ export class E2BSubprocessHandle implements SubprocessHandle {
}
private async terminateRemote(): Promise<void> {
let handle: CommandHandle | undefined
try {
handle = await this.readyState.promise
} catch {
handle = this.commandHandle
}
const handle = await this.commandState.promise
if (handle === undefined) return
const sandbox = await this.runtime.getSandbox()
const processGroupId = this.remotePid > 0 ? this.remotePid : handle.pid
this.terminationSignal = 'SIGTERM'
await this.signalGroup(sandbox, processGroupId, 'TERM')
const deadline = Date.now() + this.spec.graceMs
while (Date.now() < deadline && await this.groupAlive(sandbox, processGroupId)) {
await waitTick()
if (!isValidProcessId(handle.pid) && this.remotePid <= 0) {
await handle.kill()
this.invalidHandleQuiescent = true
this.commandHandle = undefined
return
}
if (!await this.groupAlive(sandbox, processGroupId)) return
this.terminationSignal = 'SIGKILL'
if (this.remotePid <= 0) {
const sandbox = await this.runtime.getSandbox()
this.terminationSignal = 'SIGTERM'
try {
const delivered = await this.signalGroup(sandbox, handle.pid, 'TERM')
if (delivered) {
const deadline = Date.now() + this.spec.graceMs
while (Date.now() < deadline && await this.groupAlive(sandbox, handle.pid)) await waitTick()
if (!await this.groupAlive(sandbox, handle.pid)) {
this.provisionalHandleQuiescent = true
return
}
}
} catch (_gracefulTerminationFailure) {
// A missing or unobservable provisional group still has the SDK handle fallback.
}
this.terminationSignal = 'SIGKILL'
let groupDelivered = false
let groupFailure: unknown
try {
groupDelivered = await this.signalGroup(sandbox, handle.pid, 'KILL')
} catch (error: unknown) {
groupFailure = error
}
let handleFailure: unknown
try {
await handle.kill()
this.provisionalHandleQuiescent = true
} catch (error: unknown) {
handleFailure = error
}
if (!groupDelivered && handleFailure !== undefined) {
throw new AggregateError(
[...(groupFailure === undefined ? [] : [groupFailure]), handleFailure],
'subprocess-e2b: force termination failed through both process-group and SDK transports',
)
}
return
}
const sandbox = await this.runtime.getSandbox()
const processGroupId = this.remotePid
this.terminationSignal = 'SIGTERM'
try {
await this.signalGroup(sandbox, processGroupId, 'KILL')
} finally {
await handle.kill().catch(() => false)
await this.signalGroup(sandbox, processGroupId, 'TERM')
const deadline = Date.now() + this.spec.graceMs
while (Date.now() < deadline && await this.groupAlive(sandbox, processGroupId)) {
await waitTick()
}
if (!await this.groupAlive(sandbox, processGroupId)) return
} catch (_gracefulTerminationFailure) {
// Failed TERM delivery or observation cannot prove exit; force cleanup still owns the group.
}
this.terminationSignal = 'SIGKILL'
let groupFailure: unknown
let groupDelivered = false
try {
groupDelivered = await this.signalGroup(sandbox, processGroupId, 'KILL')
} catch (error: unknown) {
groupFailure = error
}
let handleFailure: unknown
try {
await handle.kill()
} catch (error: unknown) {
handleFailure = error
}
if (!groupDelivered && handleFailure !== undefined && await this.groupAlive(sandbox, processGroupId)) {
throw new AggregateError(
[...(groupFailure === undefined ? [] : [groupFailure]), handleFailure],
'subprocess-e2b: force termination failed through both process-group and SDK transports',
)
}
}
private async signalGroup(sandbox: Sandbox, pid: number, signal: 'TERM' | 'KILL'): Promise<void> {
private throwTerminationFailure(): void {
if (this.terminationFailure !== undefined) throw this.terminationFailure
}
private async signalGroup(sandbox: Sandbox, pid: number, signal: 'TERM' | 'KILL'): Promise<boolean> {
try {
await sandbox.commands.run(`kill -${signal} -- -${pid}`)
return true
} catch (error: unknown) {
if (!(error instanceof CommandExitError)) throw error
if (error instanceof CommandExitError) return false
throw error
}
}
@@ -482,7 +656,7 @@ export class E2BSubprocessHandle implements SubprocessHandle {
if (!hasSpill(mode)) return
// A spill mode is a collect mode, so construction always created its reader.
const size = (reader as E2BOutputReader).size
if (size <= mode.maxBytes || size > mode.spill.maxBytes) {
if (this.outputDrainExpired || size <= mode.maxBytes || size > mode.spill.maxBytes) {
removals.push(sandbox.files.remove(path).catch(() => {}))
}
}
@@ -490,4 +664,18 @@ export class E2BSubprocessHandle implements SubprocessHandle {
collect(this.spec.stdio.stderr, this.stderrReader, this.paths.stderr)
await Promise.all(removals)
}
private async removeFailedState(sandbox: Sandbox): Promise<void> {
const failures: Error[] = []
for (const path of [this.paths.environment, this.stateDir]) {
try {
await sandbox.files.remove(path)
} catch (error: unknown) {
if (!(error instanceof FileNotFoundError)) failures.push(asError(error))
}
}
if (failures.length > 0) {
throw new AggregateError(failures, 'subprocess-e2b: failed to remove private command state')
}
}
}

47
packages/e2b/subprocess-e2b/src/terminal.ts
View File

@@ -376,25 +376,26 @@ export async function spawnE2BTerminal(
argv: posix.join(stateDir, 'argv'),
ready: posix.join(stateDir, 'ready'),
}
const ambient = await sandbox.commands.run('env -0', signalOpts(spec.signal))
const environment = remoteEnvironment(ambient.stdout, spec.env)
const argv = serializeValues(spec.argv, 'argv')
await sandbox.files.makeDir(stateDir)
await sandbox.commands.run(`chmod 700 -- ${quoteE2BShellArg(stateDir)}`, signalOpts(spec.signal))
await sandbox.files.write([
{ path: paths.runner, data: TERMINAL_RUNNER_SOURCE },
{ path: paths.environment, data: environment },
{ path: paths.argv, data: argv },
], signalOpts(spec.signal))
await sandbox.commands.run(
`chmod 600 -- ${quoteE2BShellArg(paths.runner)} ${quoteE2BShellArg(paths.environment)} ${quoteE2BShellArg(paths.argv)}`,
signalOpts(spec.signal),
)
const output = new PassThrough()
let handle: CommandHandle | undefined
let completion: Promise<CommandResult> | undefined
let stateDirectoryCreated = false
try {
const ambient = await sandbox.commands.run('env -0', signalOpts(spec.signal))
const environment = remoteEnvironment(ambient.stdout, spec.env)
const argv = serializeValues(spec.argv, 'argv')
await sandbox.files.makeDir(stateDir)
stateDirectoryCreated = true
await sandbox.commands.run(`chmod 700 -- ${quoteE2BShellArg(stateDir)}`, signalOpts(spec.signal))
await sandbox.files.write([
{ path: paths.runner, data: TERMINAL_RUNNER_SOURCE },
{ path: paths.environment, data: environment },
{ path: paths.argv, data: argv },
], signalOpts(spec.signal))
await sandbox.commands.run(
`chmod 600 -- ${quoteE2BShellArg(paths.runner)} ${quoteE2BShellArg(paths.environment)} ${quoteE2BShellArg(paths.argv)}`,
signalOpts(spec.signal),
)
handle = await sandbox.pty.create({
rows: spec.rows,
cols: spec.cols,
@@ -425,19 +426,25 @@ export async function spawnE2BTerminal(
)
} catch (error: unknown) {
output.destroy()
let cleanupError: Error | undefined
const cleanupErrors: Error[] = []
if (handle !== undefined && completion !== undefined) {
try {
await rollbackUnpublishedTerminal(sandbox, handle, completion, spec.graceMs)
} catch (rollbackError: unknown) {
cleanupError = asError(rollbackError)
cleanupErrors.push(asError(rollbackError))
}
} else if (handle !== undefined) {
await handle.kill().catch(() => false)
}
await sandbox.files.remove(stateDir).catch(() => {})
if (cleanupError !== undefined) {
throw new AggregateError([asError(error), cleanupError], asError(error).message)
if (stateDirectoryCreated) {
try {
await sandbox.files.remove(stateDir)
} catch (stateError: unknown) {
if (!(stateError instanceof FileNotFoundError)) cleanupErrors.push(asError(stateError))
}
}
if (cleanupErrors.length > 0) {
throw new AggregateError([asError(error), ...cleanupErrors], asError(error).message)
}
throw error
}

311
packages/e2b/subprocess-e2b/tests/subprocess.spec.ts
View File

@@ -2,6 +2,7 @@ import { once } from 'node:events'
import { Context } from 'cordis'
import {
CommandExitError,
FileNotFoundError,
type CommandHandle,
type CommandResult,
type Sandbox,
@@ -35,10 +36,14 @@ class FakeCommandHandle {
readonly sent: Array<string | Uint8Array> = []
closes = 0
kills = 0
disconnects = 0
killError: unknown
disconnectError: unknown
private readonly result = Promise.withResolvers<CommandResult>()
private settled = false
constructor(private readonly onKill: () => void = () => {}) {}
wait(): Promise<CommandResult> {
return this.result.promise
}
@@ -54,9 +59,15 @@ class FakeCommandHandle {
async kill(): Promise<boolean> {
this.kills += 1
if (this.killError !== undefined) throw this.killError
this.onKill()
return true
}
async disconnect(): Promise<void> {
this.disconnects += 1
if (this.disconnectError !== undefined) throw this.disconnectError
}
succeed(exitCode = 0): void {
if (this.settled) return
this.settled = true
@@ -77,7 +88,7 @@ class FakeCommandHandle {
}
class FakeSandbox {
readonly handle = new FakeCommandHandle()
readonly handle: FakeCommandHandle
readonly commandsSeen: string[] = []
readonly writtenFiles: string[][] = []
readonly writtenFileData = new Map<string, string>()
@@ -85,19 +96,35 @@ class FakeSandbox {
readonly directories: string[] = []
startOptions: StartOptions | undefined
backgroundError: unknown
envError: unknown
nextRemoveError: unknown
probeError: unknown
signalError: unknown
readonly signalErrors: unknown[] = []
trapsTerm = false
delaysKill = false
sdkKillStops = true
alive = true
ambient = 'PATH=/ambient/bin\0KEEP=safe\0NPM_TOKEN=secret\0DSH_STALE=old\0BROKEN\0=bad\0'
processGroupId = '4242\n'
exitStatus = ''
readonly processGroupReads: string[] = []
afterStatusRead: (() => void) | undefined
beforeProbe: (() => void) | undefined
afterProbe: (() => void) | undefined
private startGate: Promise<void> | undefined
private openStart: (() => void) | undefined
private processGroupReadGate: Promise<void> | undefined
private openProcessGroupRead: (() => void) | undefined
constructor() {
this.handle = new FakeCommandHandle(() => {
if (this.sdkKillStops) {
this.alive = false
this.handle.fail(137)
}
})
}
deferStart(): void {
const gate = Promise.withResolvers<undefined>()
@@ -109,6 +136,16 @@ class FakeSandbox {
this.openStart?.()
}
deferProcessGroupRead(): void {
const gate = Promise.withResolvers<undefined>()
this.processGroupReadGate = gate.promise
this.openProcessGroupRead = () => { gate.resolve(undefined) }
}
releaseProcessGroupRead(): void {
this.openProcessGroupRead?.()
}
finish(exitCode = 0): void {
this.alive = false
void this.completeOutput().then(
@@ -155,7 +192,14 @@ class FakeSandbox {
for (const file of files) this.writtenFileData.set(file.path, file.data)
return files.map(() => ({}))
},
read: async (): Promise<string> => this.processGroupReads.shift() ?? this.processGroupId,
read: async (path: string): Promise<string> => {
if (!path.endsWith('/exit-code')) {
await this.processGroupReadGate
return this.processGroupReads.shift() ?? this.processGroupId
}
this.afterStatusRead?.()
return this.exitStatus
},
remove: async (path: string): Promise<void> => {
this.removed.push(path)
if (this.nextRemoveError !== undefined) {
@@ -168,7 +212,10 @@ class FakeSandbox {
commands: {
run: async (command: string, options?: StartOptions | { signal?: AbortSignal }): Promise<CommandHandle | CommandResult> => {
this.commandsSeen.push(command)
if (command === 'env -0') return { exitCode: 0, stdout: this.ambient, stderr: '' }
if (command === 'env -0') {
if (this.envError !== undefined) throw this.envError
return { exitCode: 0, stdout: this.ambient, stderr: '' }
}
if (command.startsWith('kill -0 ')) {
this.beforeProbe?.()
if (options?.signal?.aborted === true) throw new DOMException('aborted', 'AbortError')
@@ -182,9 +229,9 @@ class FakeSandbox {
return { exitCode: 0, stdout: '', stderr: '' }
}
if (command.startsWith('kill -TERM ')) {
if (this.signalError !== undefined) {
const error = this.signalError
this.signalError = undefined
const error = this.signalErrors.shift() ?? this.signalError
if (error !== undefined) {
if (this.signalErrors.length === 0) this.signalError = undefined
throw error
}
if (!this.trapsTerm) {
@@ -194,9 +241,9 @@ class FakeSandbox {
return { exitCode: 0, stdout: '', stderr: '' }
}
if (command.startsWith('kill -KILL ')) {
if (this.signalError !== undefined) {
const error = this.signalError
this.signalError = undefined
const error = this.signalErrors.shift() ?? this.signalError
if (error !== undefined) {
if (this.signalErrors.length === 0) this.signalError = undefined
throw error
}
if (!this.delaysKill) this.alive = false
@@ -277,6 +324,8 @@ describe('E2BOutputReader', () => {
expect(reader.readFrom(2)).toEqual({ text: 'cdef', nextOffset: 6, lossy: false })
expect(reader.readFrom(5)).toEqual({ text: 'f', nextOffset: 6, lossy: false })
expect(reader.readFrom(99)).toEqual({ text: '', nextOffset: 6, lossy: false })
reader.invalidateSpill()
expect(reader.readFrom(0)).toEqual({ text: 'cdef', nextOffset: 6, lossy: true })
})
it('drops whole head chunks and withholds absent or over-cap spills', () => {
@@ -390,6 +439,50 @@ describe('E2BSubprocessHandle', () => {
await expect(handle.done).rejects.toThrow('incomplete output transport')
})
it('bounds descendant-held output draining and withholds the incomplete spill', async () => {
const fake = new FakeSandbox()
const handle = new E2BSubprocessHandle(runtime(fake), spec({ graceMs: 5 }), '/runtime/drain-bound')
await flush()
await fake.stdout('leader-output')
fake.exitStatus = '0\n'
await expect(handle.done).resolves.toEqual({ exitCode: 0, signal: null })
expect(fake.handle.disconnects).toBe(1)
expect(handle.collected.stdout?.readFrom(0)).toEqual({
text: 'tput',
nextOffset: 13,
lossy: true,
})
expect(fake.removed).toContain('/runtime/drain-bound/stdout.log')
handle.terminate()
await expect(handle.waitForExit()).resolves.toBe(true)
})
it('accepts clean encoder completion inside the output-drain grace', async () => {
const fake = new FakeSandbox()
const handle = new E2BSubprocessHandle(runtime(fake), spec({ graceMs: 100 }), '/runtime/drain-complete')
await flush()
fake.exitStatus = '0\n'
fake.afterStatusRead = () => {
fake.afterStatusRead = undefined
setTimeout(() => { fake.finish() }, 0)
}
await expect(handle.done).resolves.toEqual({ exitCode: 0, signal: null })
expect(fake.handle.disconnects).toBe(0)
})
it('rejects an invalid direct-command exit status', async () => {
const fake = new FakeSandbox()
const handle = new E2BSubprocessHandle(runtime(fake), spec(), '/runtime/invalid-status')
await flush()
fake.exitStatus = '999\n'
await expect(handle.done).rejects.toThrow('invalid exit code')
handle.terminate()
await expect(handle.waitForExit()).resolves.toBe(true)
})
it('surfaces deferred piped-stdin write and close failures as stream errors', async () => {
const writeFake = new FakeSandbox()
writeFake.deferStart()
@@ -510,6 +603,97 @@ describe('E2BSubprocessHandle', () => {
await expect(handle.done).resolves.toEqual({ exitCode: null, signal: 'SIGTERM' })
})
it('kills through the provisional SDK handle before process-group publication', async () => {
const fake = new FakeSandbox()
fake.deferProcessGroupRead()
fake.signalErrors.push(commandError(1), commandError(1))
const handle = new E2BSubprocessHandle(runtime(fake), spec(), '/runtime/pre-publication-kill')
await vi.waitFor(() => { expect(fake.startOptions).toBeDefined() })
handle.terminate()
await vi.waitFor(() => { expect(fake.handle.kills).toBe(1) })
expect(fake.alive).toBe(false)
await expect(handle.waitForExit()).resolves.toBe(true)
fake.releaseProcessGroupRead()
await expect(handle.done).resolves.toEqual({ exitCode: null, signal: 'SIGKILL' })
})
it('bounds a quiescence observer while provisional termination is awaiting the controller', async () => {
const fake = new FakeSandbox()
fake.deferProcessGroupRead()
const reconnect = Promise.withResolvers<Sandbox>()
let calls = 0
const delayedRuntime = runtime(fake, async () => {
calls += 1
return calls === 1 ? fake.sandbox : await reconnect.promise
})
const handle = new E2BSubprocessHandle(delayedRuntime, spec(), '/runtime/pre-publication-observer')
await vi.waitFor(() => { expect(fake.startOptions).toBeDefined() })
handle.terminate()
const controller = new AbortController()
const waiting = handle.waitForExit(controller.signal)
await flush()
controller.abort()
await expect(waiting).resolves.toBe(false)
reconnect.resolve(fake.sandbox)
await expect(handle.waitForExit()).resolves.toBe(true)
fake.releaseProcessGroupRead()
await handle.done
})
it('proves a provisional group exit when the SDK kill fallback fails', async () => {
const fake = new FakeSandbox()
fake.deferProcessGroupRead()
fake.trapsTerm = true
fake.handle.killError = new Error('SDK kill unavailable')
const handle = new E2BSubprocessHandle(runtime(fake), spec({ graceMs: 1 }), '/runtime/pre-publication-group-kill')
await vi.waitFor(() => { expect(fake.startOptions).toBeDefined() })
handle.terminate()
await expect(handle.waitForExit()).resolves.toBe(true)
fake.releaseProcessGroupRead()
await expect(handle.done).resolves.toEqual({ exitCode: null, signal: 'SIGKILL' })
})
it('reports failed provisional group and SDK force transports', async () => {
const fake = new FakeSandbox()
fake.deferProcessGroupRead()
fake.signalErrors.push(new Error('TERM transport failed'), new Error('KILL transport failed'))
fake.handle.killError = new Error('SDK kill failed')
const handle = new E2BSubprocessHandle(runtime(fake), spec({ graceMs: 1 }), '/runtime/pre-publication-failure')
await vi.waitFor(() => { expect(fake.startOptions).toBeDefined() })
handle.terminate()
await expect(handle.waitForExit()).rejects.toThrow('force termination failed through both')
fake.handle.killError = undefined
handle.terminate()
await expect(handle.waitForExit()).resolves.toBe(true)
fake.releaseProcessGroupRead()
await handle.done
const absentGroup = new FakeSandbox()
absentGroup.deferProcessGroupRead()
absentGroup.signalErrors.push(commandError(1), commandError(1))
absentGroup.handle.killError = new Error('SDK kill failed without a provisional group')
const absentHandle = new E2BSubprocessHandle(
runtime(absentGroup),
spec({ graceMs: 1 }),
'/runtime/pre-publication-absent-group',
)
await vi.waitFor(() => { expect(absentGroup.startOptions).toBeDefined() })
absentHandle.terminate()
await expect(absentHandle.waitForExit()).rejects.toThrow('force termination failed through both')
absentGroup.handle.killError = undefined
absentHandle.terminate()
await expect(absentHandle.waitForExit()).resolves.toBe(true)
absentGroup.releaseProcessGroupRead()
await absentHandle.done
})
it('honors an already-aborted signal when constructing the asynchronous handle directly', async () => {
const fake = new FakeSandbox()
const handle = new E2BSubprocessHandle(runtime(fake), spec({ signal: AbortSignal.abort('stop') }), '/runtime/pre-aborted')
@@ -559,6 +743,17 @@ describe('E2BSubprocessHandle', () => {
await expect(liveWait).resolves.toBe(false)
fake.finish()
await handle.done
const terminatingFake = new FakeSandbox()
terminatingFake.deferStart()
const terminating = new E2BSubprocessHandle(runtime(terminatingFake), spec(), '/runtime/wait-termination-start')
terminating.terminate()
const beforeHandle = new AbortController()
const handlePending = terminating.waitForExit(beforeHandle.signal)
beforeHandle.abort()
await expect(handlePending).resolves.toBe(false)
terminatingFake.releaseStart()
await terminating.done
})
it('bounds both sides of the liveness-poll abort race', async () => {
@@ -580,6 +775,14 @@ describe('E2BSubprocessHandle', () => {
const duringProbe = new AbortController()
fake.beforeProbe = () => { duringProbe.abort(); fake.beforeProbe = undefined }
await expect(handle.waitForExit(duringProbe.signal)).resolves.toBe(false)
let racedAbort = false
const raceSignal = {
get aborted() { return racedAbort },
addEventListener: () => { racedAbort = true },
removeEventListener: () => {},
} as unknown as AbortSignal
await expect(handle.waitForExit(raceSignal)).resolves.toBe(false)
fake.finish()
await handle.done
})
@@ -599,8 +802,36 @@ describe('E2BSubprocessHandle', () => {
const handle = new E2BSubprocessHandle(runtime(fake), spec(), '/runtime/fail')
await expect(handle.done).rejects.toThrow('start failed')
expect(handle.pid).toBe(-1)
expect(fake.removed).toContain('/runtime/fail/environment')
expect(fake.removed).toContain('/runtime/fail')
await expect(handle.waitForExit()).resolves.toBe(true)
handle.terminate()
const unavailableHandle = new E2BSubprocessHandle(
runtime(new FakeSandbox(), async () => { throw new Error('sandbox unavailable') }),
spec(),
'/runtime/unavailable-start',
)
await expect(unavailableHandle.done).rejects.toThrow('sandbox unavailable')
await expect(unavailableHandle.waitForExit()).resolves.toBe(true)
const envFailure = new FakeSandbox()
envFailure.envError = new Error('ambient lookup failed')
const envHandle = new E2BSubprocessHandle(runtime(envFailure), spec(), '/runtime/env-failure')
await expect(envHandle.done).rejects.toThrow('ambient lookup failed')
expect(envFailure.removed).toEqual([])
const cleanupFailure = new FakeSandbox()
cleanupFailure.backgroundError = new Error('start failed before credential consumption')
cleanupFailure.nextRemoveError = new Error('credential cleanup failed')
const cleanupHandle = new E2BSubprocessHandle(runtime(cleanupFailure), spec(), '/runtime/cleanup-failure')
await expect(cleanupHandle.done).rejects.toThrow('command failed and private state cleanup failed')
const absentState = new FakeSandbox()
absentState.backgroundError = new Error('start failed after external cleanup')
absentState.nextRemoveError = new FileNotFoundError('already removed')
const absentHandle = new E2BSubprocessHandle(runtime(absentState), spec(), '/runtime/absent-state')
await expect(absentHandle.done).rejects.toThrow('start failed after external cleanup')
})
it('bounds a readiness rejection with a still-live caller signal', async () => {
@@ -678,8 +909,20 @@ describe('E2BSubprocessHandle', () => {
invalidPid.handle.pid = 0
const invalid = new E2BSubprocessHandle(runtime(invalidPid), spec(), '/runtime/invalid-pid')
await expect(invalid.done).rejects.toThrow(/invalid command pid 0/)
expect(invalidPid.handle.kills).toBe(1)
expect(invalidPid.removed).toContain('/runtime/invalid-pid/environment')
await expect(invalid.waitForExit()).resolves.toBe(true)
const failedRollback = new FakeSandbox()
failedRollback.handle.pid = 0
failedRollback.handle.killError = new Error('invalid handle kill failed')
const retained = new E2BSubprocessHandle(runtime(failedRollback), spec(), '/runtime/invalid-pid-retained')
await expect(retained.done).rejects.toThrow('invalid command pid rollback did not reach quiescence')
await expect(retained.waitForExit()).rejects.toThrow('invalid handle kill failed')
failedRollback.handle.killError = undefined
retained.terminate()
await expect(retained.waitForExit()).resolves.toBe(true)
const crashedFake = new FakeSandbox()
const crashed = new E2BSubprocessHandle(runtime(crashedFake), spec(), '/runtime/crashed')
await flush()
@@ -692,6 +935,7 @@ describe('E2BSubprocessHandle', () => {
const invalidGroup = new FakeSandbox()
invalidGroup.processGroupId = 'not-a-pid\n'
invalidGroup.delaysKill = true
invalidGroup.sdkKillStops = false
invalidGroup.afterProbe = () => { invalidGroup.alive = false }
const invalid = new E2BSubprocessHandle(runtime(invalidGroup), spec(), '/runtime/invalid-group')
await expect(invalid.done).rejects.toThrow(/invalid process-group id/)
@@ -791,7 +1035,7 @@ describe('E2BSubprocessHandle', () => {
await expect(handle.done).resolves.toEqual({ exitCode: 0, signal: null })
})
it('contains an already-gone group signal and observes non-command signal failures', async () => {
it('contains an already-gone group signal and escalates after a TERM transport failure', async () => {
const gone = new FakeSandbox()
gone.trapsTerm = true
gone.signalError = commandError(1)
@@ -805,9 +1049,35 @@ describe('E2BSubprocessHandle', () => {
const failedHandle = new E2BSubprocessHandle(runtime(failed), spec(), '/runtime/failed-signal')
await flush()
failedHandle.terminate()
await expect(failedHandle.done).resolves.toEqual({ exitCode: null, signal: 'SIGKILL' })
expect(failed.commandsSeen).toContain('kill -KILL -- -4242')
})
it('allows termination retry after both force transports fail', async () => {
const fake = new FakeSandbox()
fake.signalErrors.push(new Error('TERM transport failed'), new Error('KILL transport failed'))
fake.handle.killError = new Error('SDK kill failed')
const handle = new E2BSubprocessHandle(runtime(fake), spec({ graceMs: 1 }), '/runtime/retry-signal')
await flush()
failed.finish()
await expect(failedHandle.done).resolves.toEqual({ exitCode: 0, signal: null })
handle.terminate()
await expect(handle.waitForExit()).rejects.toThrow('force termination failed through both')
fake.handle.killError = undefined
handle.terminate()
await expect(handle.done).resolves.toEqual({ exitCode: null, signal: 'SIGTERM' })
expect(fake.commandsSeen.filter(command => command.startsWith('kill -TERM '))).toHaveLength(2)
const missingGroup = new FakeSandbox()
missingGroup.trapsTerm = true
missingGroup.signalErrors.push(undefined, commandError(1))
missingGroup.handle.killError = new Error('SDK kill failed after group exit race')
const raced = new E2BSubprocessHandle(runtime(missingGroup), spec({ graceMs: 1 }), '/runtime/group-exit-race')
await flush()
raced.terminate()
await expect(raced.waitForExit()).rejects.toThrow('force termination failed through both')
missingGroup.handle.killError = undefined
raced.terminate()
await expect(raced.waitForExit()).resolves.toBe(true)
})
})
@@ -833,6 +1103,23 @@ describe('E2BSubprocessService', () => {
expect(fake.alive).toBe(false)
})
it('reports a failed termination transaction from disposal instead of waiting on done', async () => {
const fake = new FakeSandbox()
fake.signalErrors.push(new Error('TERM transport failed'), new Error('KILL transport failed'))
fake.handle.killError = new Error('SDK kill failed')
const { ctx, fiber } = await service(fake)
const handle = ctx.subprocess.spawn(spec({ graceMs: 1 }))
await flush()
await expect(fiber.dispose()).resolves.toBeUndefined()
await expect(handle.waitForExit()).rejects.toThrow('force termination failed through both')
fake.handle.killError = undefined
handle.terminate()
await expect(handle.waitForExit()).resolves.toBe(true)
await expect(handle.done).resolves.toEqual({ exitCode: null, signal: 'SIGTERM' })
})
it('releases naturally settled handles before later service disposal', async () => {
const fake = new FakeSandbox()
const { ctx, fiber } = await service(fake)

16
packages/e2b/subprocess-e2b/tests/terminal.spec.ts
View File

@@ -95,6 +95,7 @@ class FakeTerminalSandbox {
foreground = '456\n'
groups = [123]
createError: unknown
writeError: unknown
sendError: unknown
commandFailure: unknown
sessionGroupsFailure: unknown
@@ -116,6 +117,7 @@ class FakeTerminalSandbox {
},
write: async (files: Array<{ path: string; data: string }>): Promise<object[]> => {
for (const file of files) this.writes.set(file.path, file.data)
if (this.writeError !== undefined) throw this.writeError
return files.map(() => ({}))
},
read: async (): Promise<string> => {
@@ -278,6 +280,20 @@ describe('E2B terminal allocation', () => {
})
it('cleans malformed handles, bootstrap failures, and readiness failures', async () => {
const failedState = new FakeTerminalSandbox()
failedState.writeError = new Error('state write failed')
await expect(spawnE2BTerminal(runtime(failedState), spec(), '/runtime/state-write'))
.rejects.toThrow('state write failed')
expect(failedState.writes.get('/runtime/state-write/environment')).toContain('KEEP=visible\0')
expect(failedState.removed).toContain('/runtime/state-write')
expect(failedState.createOptions).toBeUndefined()
const stateAlreadyGone = new FakeTerminalSandbox()
stateAlreadyGone.writeError = new Error('state write failed after external cleanup')
stateAlreadyGone.removeError = new FileNotFoundError('state already gone')
await expect(spawnE2BTerminal(runtime(stateAlreadyGone), spec(), '/runtime/state-gone'))
.rejects.toThrow('state write failed after external cleanup')
const invalidPid = new FakeTerminalSandbox()
invalidPid.handle.pid = 0
await expect(spawnE2BTerminal(runtime(invalidPid), spec(), '/runtime/invalid-pid'))