Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
TypeScript
import { describe, expect, it, vi } from 'vitest'
import { EventEmitter } from 'node:events'
import { ExecOutputLimitError, K8sGateway } from '../src/k8s.js'
let capturedStdout: { destroyed: boolean; write: (chunk: Buffer) => boolean } | null = null
let capturedStderr: { destroyed: boolean } | null = null
let closeSpy: ReturnType<typeof vi.fn> | null = null
vi.mock('@kubernetes/client-node', () => ({
Exec: vi.fn().mockImplementation(function ExecMock() {
return {
exec: vi.fn(async (_namespace, _podName, _containerName, _command, stdout, stderr) => {
capturedStdout = stdout
capturedStderr = stderr
const conn = new EventEmitter() as EventEmitter & { close: ReturnType<typeof vi.fn> }
closeSpy = vi.fn(() => conn.emit('close'))
conn.close = closeSpy
return conn
}),
}
}),
}))
describe('K8sGateway execBytes stream lifecycle', () => {
it('rejects with a typed limit error, destroys streams, and closes the connection on overflow', async () => {
capturedStdout = null
capturedStderr = null
closeSpy = null
const gateway = Object.create(K8sGateway.prototype) as K8sGateway
Object.defineProperty(gateway, 'kc', { value: {} })
const result = (
gateway as unknown as {
execBytes: (
podName: string,
namespace: string,
containerName: string | undefined,
command: string[],
maxBytes: number,
maxBytesExceededMessage: string
) => Promise<Buffer>
}
).execBytes(
'pod-1',
'mcp-host',
undefined,
['cat', '/tmp/clerum-output/report.md'],
4,
'Artifact listing too large to return'
)
await vi.waitFor(() => expect(capturedStdout).not.toBeNull())
capturedStdout!.write(Buffer.alloc(5))
await expect(result).rejects.toBeInstanceOf(ExecOutputLimitError)
expect(capturedStdout!.destroyed).toBe(true)
expect(capturedStderr!.destroyed).toBe(true)
expect(closeSpy).toHaveBeenCalledTimes(1)
})
})