Files
kandev/apps/web/e2e/tests/ssh/launch-task.spec.ts
T
José Almeida d744a4e60c test(e2e): align H5 assertion with actual on-disk session-dir layout
The wrapper writes agentctl.pid and agentctl.log to <sessionDir>; the
remote agentctl port travels via AGENTCTL_PORT env + ExecutorRunning
metadata, not a separate .port file. Assert the files that actually
exist and the metadata field that carries the port.
2026-05-17 23:45:07 +01:00

202 lines
7.4 KiB
TypeScript
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
import { test, expect } from "../../fixtures/ssh-test-base";
import { listRemoteDir, readRemoteFile, remotePathExists } from "../../helpers/ssh";
import { waitForLatestSessionDone } from "../../helpers/session";
/**
* Full end-to-end task launch on the real sshd container. The smoke test
* for the SSH executor: upload agentctl, mkdir the per-task dir, clone the
* repo, launch the per-session agentctl, port-forward, run the agent to
* completion, observe the on-remote filesystem layout.
*
* Covers e2e-plan.md group H (H1–H8).
*/
test.describe("ssh executor — task launch", () => {
test("launches a session and records ssh runtime on the task environment", async ({
apiClient,
seedData,
}) => {
test.setTimeout(180_000);
const task = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H1 launch",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, task.id, 1, "Wait for first SSH session");
const env = await apiClient.getTaskEnvironment(task.id);
expect(env).not.toBeNull();
expect(env!.executor_type).toBe("ssh");
});
test("agentctl is uploaded on first launch and sha256 sidecar lands", async ({
apiClient,
seedData,
}) => {
test.setTimeout(180_000);
const task = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H2 upload agentctl",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, task.id, 1, "Wait for upload");
expect(remotePathExists(seedData.sshTarget, "/home/kandev/.kandev/bin/agentctl")).toBe(true);
expect(remotePathExists(seedData.sshTarget, "/home/kandev/.kandev/bin/agentctl.sha256")).toBe(
true,
);
const sha = readRemoteFile(seedData.sshTarget, "/home/kandev/.kandev/bin/agentctl.sha256");
expect(sha.trim()).toMatch(/^[0-9a-f]{64}$/);
});
test("second launch on the same host skips the upload (sha matches)", async ({
apiClient,
seedData,
}) => {
test.setTimeout(240_000);
// First launch primes the cache.
const first = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H3 first launch",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, first.id, 1, "First launch");
// Inspect mtime BEFORE the second launch.
const beforeMtime = readRemoteFile(
seedData.sshTarget,
"/home/kandev/.kandev/bin/agentctl.sha256",
);
const second = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H3 second launch",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, second.id, 1, "Second launch");
// The sha256 file should be byte-identical (same hash, content reused).
const afterMtime = readRemoteFile(
seedData.sshTarget,
"/home/kandev/.kandev/bin/agentctl.sha256",
);
expect(afterMtime).toBe(beforeMtime);
});
test("per-task workdir lives at <workdir>/tasks/<task-dir>/", async ({ apiClient, seedData }) => {
test.setTimeout(180_000);
const task = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H4 task workdir",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, task.id, 1, "Wait for workdir");
const sessions = await apiClient.listSSHSessions(seedData.sshExecutorId);
const row = sessions.find((s) => s.task_id === task.id);
expect(row?.remote_task_dir).toMatch(/\/tasks\/[^/]+$/);
expect(remotePathExists(seedData.sshTarget, row!.remote_task_dir!)).toBe(true);
});
test("per-session runtime dir holds the agentctl pid and port files", async ({
apiClient,
seedData,
}) => {
test.setTimeout(180_000);
const task = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H5 session runtime",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, task.id, 1, "Wait for session runtime");
const sessions = await apiClient.listSSHSessions(seedData.sshExecutorId);
const row = sessions.find((s) => s.task_id === task.id);
expect(row).toBeDefined();
const sessionDir = `${row!.remote_task_dir}/.kandev/sessions/${row!.session_id}`;
expect(remotePathExists(seedData.sshTarget, sessionDir)).toBe(true);
const entries = listRemoteDir(seedData.sshTarget, sessionDir);
// The wrapper writes the pid file and the log; the port travels via the
// AGENTCTL_PORT env var and the ExecutorRunning metadata, not a file.
expect(entries).toEqual(expect.arrayContaining(["agentctl.pid", "agentctl.log"]));
expect(row!.remote_agentctl_port).toBeGreaterThan(0);
});
test("stopping the session cleans up the session runtime dir but leaves the task dir", async ({
apiClient,
seedData,
}) => {
test.setTimeout(180_000);
const task = await apiClient.createTaskWithAgent(
seedData.workspaceId,
"H7 stop cleanup",
seedData.agentProfileId,
{
description: "/e2e:simple-message",
workflow_id: seedData.workflowId,
workflow_step_id: seedData.startStepId,
repository_ids: [seedData.repositoryId],
executor_profile_id: seedData.sshExecutorProfileId,
},
);
await waitForLatestSessionDone(apiClient, task.id, 1, "Wait before stop");
const sessions = await apiClient.listSSHSessions(seedData.sshExecutorId);
const row = sessions.find((s) => s.task_id === task.id);
expect(row).toBeDefined();
const sessionDir = `${row!.remote_task_dir}/.kandev/sessions/${row!.session_id}`;
// Archive (or otherwise end) the task to trigger StopInstance.
await apiClient.archiveTask(task.id);
await expect
.poll(() => remotePathExists(seedData.sshTarget, sessionDir), { timeout: 30_000 })
.toBe(false);
// Task dir intact — v1 spec, no auto-clean on last session stop.
expect(remotePathExists(seedData.sshTarget, row!.remote_task_dir!)).toBe(true);
});
});