Ryan Malloy
f053762c4e
The success paths were fine; several failure paths drew a confident conclusion without checking the thing they waited for. Registry (was: any parse error killed the whole server, since load() runs in the lifespan): - quarantine an unreadable file and start empty instead of raising, so the tools that stop runaway VMs keep working when bookkeeping is damaged - skip malformed or invalidly-named records rather than failing the load; report both through list_vms as registry_warnings - read-modify-write under an exclusive flock so a second instance merges instead of clobbering, with a PID-unique temp file - drop the lifespan shutdown save, which could resurrect deleted records - version the schema and round-trip unknown record fields sandbox_destroy (the only tool that deletes files): - verify the process actually died, escalating to SIGKILL, and refuse to delete an overlay QEMU still holds open - assert the target is inside the VM state tree before rmtree - report cleanup errors instead of swallowing them; destroyed now reflects what happened Launch races: - reserve the name before the first await so two concurrent launches cannot race over one set of sockets - refuse to unlink a QMP socket that is still accepting connections - register the VM with a warning rather than orphaning it when the pidfile is unreadable but QEMU is up Guest agent and QMP: - bound every guest-agent call, not just the handshake; cap max_bytes and stop guest_file_read spinning on a zero-progress agent - serialize QMP sessions per VM (the monitor is single-client) and say "another operation holds it" instead of "the VM has likely exited" - poll liveness while waiting for SHUTDOWN so a crashed VM is reported as exited rather than as a guest ignoring ACPI - default command timeout, with a longer bound for savevm/loadvm - stricter snapshot tags; log destructive operations to stderr Adds tests/test_reliability.py covering the conditions above.
165 lines
6.8 KiB
Python
165 lines
6.8 KiB
Python
"""Lifecycle tools with mocked spawn / fake QMP."""
|
|
|
|
import pytest
|
|
from fastmcp import Client
|
|
from fastmcp.exceptions import ToolError
|
|
|
|
from conftest import FakeQMPClient, result_data, write_registry
|
|
from mcqemu.models import VMRecord
|
|
from mcqemu.server import mcp
|
|
|
|
|
|
def seeded_record(dirs, name="vm1", **kw) -> VMRecord:
|
|
defaults = dict(
|
|
source="spawned",
|
|
qmp_socket=str(dirs.run / name / "qmp.sock"),
|
|
qga_socket=str(dirs.run / name / "qga.sock"),
|
|
pid=4242,
|
|
arch="x86_64",
|
|
config={"disks": []},
|
|
)
|
|
defaults.update(kw)
|
|
return VMRecord(name=name, **defaults)
|
|
|
|
|
|
async def test_launch_rejects_bad_name(client):
|
|
with pytest.raises(ToolError, match="Invalid VM name"):
|
|
await client.call_tool("launch_vm", {"name": "bad name!"})
|
|
|
|
|
|
async def test_launch_rejects_missing_disk(client, tmp_path):
|
|
with pytest.raises(ToolError, match="image not found"):
|
|
await client.call_tool(
|
|
"launch_vm", {"name": "vm1", "disks": [str(tmp_path / "nope.qcow2")]}
|
|
)
|
|
|
|
|
|
async def test_launch_rejects_disk_of_running_vm(dirs, all_pids_alive, tmp_path, monkeypatch):
|
|
disk = tmp_path / "busy.qcow2"
|
|
disk.touch()
|
|
write_registry(dirs, seeded_record(dirs, "other", config={"disks": [str(disk)]}))
|
|
async with Client(mcp) as client:
|
|
with pytest.raises(ToolError, match="already attached to running VM 'other'"):
|
|
await client.call_tool("launch_vm", {"name": "vm2", "disks": [str(disk)]})
|
|
|
|
|
|
async def test_launch_rejects_duplicate_running_name(dirs, all_pids_alive):
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
with pytest.raises(ToolError, match="already running"):
|
|
await client.call_tool("launch_vm", {"name": "vm1"})
|
|
|
|
|
|
async def test_launch_happy_path_registers(dirs, monkeypatch):
|
|
async def fake_spawn(cfg, config):
|
|
return seeded_record(dirs, cfg.name, accel="kvm", config=cfg.__dict__.copy())
|
|
|
|
monkeypatch.setattr("mcqemu.tools.lifecycle.spawn_vm", fake_spawn)
|
|
async with Client(mcp) as client:
|
|
data = result_data(await client.call_tool("launch_vm", {"name": "fresh", "no_net": True}))
|
|
assert data["status"] == "running"
|
|
assert data["accel"] == "kvm"
|
|
result = result_data(await client.call_tool("list_vms", {}))
|
|
assert [v["name"] for v in result["vms"]] == ["fresh"]
|
|
|
|
|
|
async def test_stop_force_sends_quit(dirs, fake_qmp, all_pids_dead, monkeypatch):
|
|
monkeypatch.setattr("mcqemu.tools.lifecycle.Path.exists", lambda self: True)
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
data = result_data(await client.call_tool("stop_vm", {"name": "vm1", "force": True}))
|
|
assert data["status"] == "stopped"
|
|
assert ("quit", {}) in FakeQMPClient.calls
|
|
|
|
|
|
async def test_stop_graceful_waits_for_shutdown_event(dirs, fake_qmp, all_pids_dead, monkeypatch):
|
|
# is_process_alive is False (all_pids_dead), so make the socket "exist"
|
|
# to get past the already-stopped short-circuit.
|
|
monkeypatch.setattr("mcqemu.tools.lifecycle.Path.exists", lambda self: True)
|
|
FakeQMPClient.events_to_emit = [{"event": "SHUTDOWN"}]
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
data = result_data(await client.call_tool("stop_vm", {"name": "vm1", "timeout": 2}))
|
|
assert data["method"] == "graceful ACPI shutdown"
|
|
assert ("system_powerdown", {}) in FakeQMPClient.calls
|
|
|
|
|
|
async def test_stop_graceful_timeout_advises_force(dirs, fake_qmp, all_pids_alive, monkeypatch):
|
|
# Guest ignores ACPI: no SHUTDOWN event AND the process stays alive.
|
|
monkeypatch.setattr("mcqemu.tools.lifecycle.Path.exists", lambda self: True)
|
|
FakeQMPClient.events_to_emit = []
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
with pytest.raises(ToolError, match="force=True"):
|
|
await client.call_tool("stop_vm", {"name": "vm1", "timeout": 1})
|
|
|
|
|
|
async def test_stop_graceful_reports_process_exit(dirs, fake_qmp, all_pids_dead, monkeypatch):
|
|
"""A VM that dies mid-shutdown must not be misreported as ignoring ACPI."""
|
|
monkeypatch.setattr("mcqemu.tools.lifecycle.Path.exists", lambda self: True)
|
|
FakeQMPClient.events_to_emit = [] # crashed before emitting SHUTDOWN
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
data = result_data(await client.call_tool("stop_vm", {"name": "vm1", "timeout": 30}))
|
|
assert data["status"] == "stopped"
|
|
assert "exited" in data["method"]
|
|
|
|
|
|
async def test_stop_already_stopped(dirs, all_pids_dead):
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
data = result_data(await client.call_tool("stop_vm", {"name": "vm1"}))
|
|
assert data["status"] == "stopped"
|
|
assert "already" in data["note"]
|
|
|
|
|
|
async def test_pause_resume(dirs, fake_qmp, all_pids_alive):
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
assert (
|
|
result_data(await client.call_tool("pause_vm", {"name": "vm1"}))["status"] == "paused"
|
|
)
|
|
assert (
|
|
result_data(await client.call_tool("resume_vm", {"name": "vm1"}))["status"] == "running"
|
|
)
|
|
assert ("stop", {}) in FakeQMPClient.calls
|
|
assert ("cont", {}) in FakeQMPClient.calls
|
|
|
|
|
|
async def test_unknown_vm_lists_known(dirs, fake_qmp):
|
|
write_registry(dirs, seeded_record(dirs, "alpha"))
|
|
async with Client(mcp) as client:
|
|
with pytest.raises(ToolError, match="alpha"):
|
|
await client.call_tool("pause_vm", {"name": "nope"})
|
|
|
|
|
|
async def test_attach_requires_existing_socket(client, tmp_path):
|
|
with pytest.raises(ToolError, match="No socket"):
|
|
await client.call_tool(
|
|
"attach_vm", {"name": "ext", "qmp_socket": str(tmp_path / "no.sock")}
|
|
)
|
|
|
|
|
|
async def test_attach_and_forget(dirs, fake_qmp, tmp_path):
|
|
sock = tmp_path / "ext.sock"
|
|
sock.touch()
|
|
FakeQMPClient.responses["query-status"] = {"status": "running"}
|
|
async with Client(mcp) as client:
|
|
data = result_data(
|
|
await client.call_tool("attach_vm", {"name": "ext", "qmp_socket": str(sock)})
|
|
)
|
|
assert data["status"] == "running"
|
|
data = result_data(await client.call_tool("forget_vm", {"name": "ext"}))
|
|
assert data["forgotten"] is True
|
|
result = result_data(await client.call_tool("list_vms", {}))
|
|
assert result["vms"] == []
|
|
|
|
|
|
async def test_forget_refuses_running_spawned_vm(dirs, all_pids_alive):
|
|
write_registry(dirs, seeded_record(dirs, "vm1"))
|
|
async with Client(mcp) as client:
|
|
with pytest.raises(ToolError, match="still running"):
|
|
await client.call_tool("forget_vm", {"name": "vm1"})
|
|
data = result_data(await client.call_tool("forget_vm", {"name": "vm1", "force": True}))
|
|
assert data["forgotten"] is True
|