linstor: test VM startup on disk failure

This adds a test to the LINSTOR SR test suite to make sure that a VM
with a VDI on a shared LINSTOR SR can still start and shut down properly
when a physical disk of that SR has failed.

The test does the following:
- Fails a physical disk of the LINSTOR SR pool on a random host.
- Ensures a VM can still start up and shut down on all hosts.

This uses a device mapper to avoid relying on the capabilities of the
underlying block device.

Signed-off-by: Alexandre Sollier <alexandre.sollier@vates.tech>

Author: Kuruyia

tests/storage/linstor/unhealthy/__init__.py

@@ -0,0 +1,99 @@
+from __future__ import annotations
+
+import pytest
+
+import json
+import logging
+
+from typing import TYPE_CHECKING, Generator
+
+if TYPE_CHECKING:
+    from lib.host import Host
+    from lib.pool import Pool
+
+DM_FLAKEY_DEV_NAME = 'linfail'
+
+class FlakeyDisk:
+    def __init__(
+        self,
+        host: Host,
+        device: Host.BlockDeviceInfo,
+        dm_dev_name: str,
+    ) -> None:
+        self._host = host
+        self._device = device
+        self._dm_dev_name = dm_dev_name
+
+    @property
+    def path(self) -> str:
+        return f'/dev/mapper/{self._dm_dev_name}'
+
+    def create(self) -> None:
+        self._host.ssh(f'dmsetup create {self._dm_dev_name} --table "{self._build_dm_table(False)}"')
+
+    def remove(self) -> None:
+        self._host.ssh(f'dmsetup remove {self._dm_dev_name}')
+
+    def fail(self) -> None:
+        logging.info(f'Failing device {self._device.path} on {self._host.hostname_or_ip}')
+
+        self._apply_dm_table(self._build_dm_table(True))
+        self._host.ssh('sync')
+        self._host.ssh('echo 3 > /proc/sys/vm/drop_caches')
+
+    def repair(self) -> None:
+        logging.info(f'Repairing device {self._device.path} on {self._host.hostname_or_ip}')
+
+        self._apply_dm_table(self._build_dm_table(False))
+        cmd_res = self._host.ssh('linstor -m --controllers `xe host-list params=address --minimal` r l -n `hostname` --props DrbdOptions/SkipDisk') # noqa: E501
+        failing_resources = json.loads(cmd_res)
+
+        # Make sure we take care of the database first so repairing the other
+        # resources doesn't hang or fail.
+        failing_resource_names = [res['name'] for res in failing_resources[0]]
+        failing_resource_names.sort(key=lambda x: x != 'xcp-persistent-database')
+
+        for failing_resource_name in failing_resource_names:
+            self._host.ssh(f'linstor --controllers `xe host-list params=address --minimal` r sp `hostname` {failing_resource_name} DrbdOptions/SkipDisk') # noqa: E501
+            self._host.ssh(f'drbdadm wait-sync {failing_resource_name}')
+
+    def _build_dm_table(self, disk_failed: bool) -> str:
+        disk_size = self._device.size // 512
+
+        if disk_failed:
+            return f'0 {disk_size} flakey {self._device.path} 0 0 1'
+        else:
+            return f'0 {disk_size} flakey {self._device.path} 0 1 0'
+
+    def _apply_dm_table(self, table: str) -> None:
+        self._host.ssh(f'dmsetup reload {self._dm_dev_name} --table "{table}"')
+        self._host.ssh(f'dmsetup resume {self._dm_dev_name}')
+
+@pytest.fixture(scope='package')
+def flakey_unused_512B_disk(
+    pool_with_unused_512B_disk: Pool,
+    unused_512B_disks: dict[Host, list[Host.BlockDeviceInfo]],
+) -> Generator[dict[Host, FlakeyDisk], None, None]:
+    flakey_disks: dict[Host, FlakeyDisk] = {}
+    hosts = pool_with_unused_512B_disk.hosts
+
+    for host in hosts:
+        disk = unused_512B_disks[host][0]
+        flakey_disk = FlakeyDisk(host, disk, DM_FLAKEY_DEV_NAME)
+        flakey_disk.create()
+
+        flakey_disks[host] = flakey_disk
+
+    yield flakey_disks
+
+    for flakey_disk in flakey_disks.values():
+        flakey_disk.remove()
+
+@pytest.fixture(scope='package')
+def lvm_disk_paths(
+    flakey_unused_512B_disk: dict[Host, FlakeyDisk],
+) -> dict[Host, list[str]]:
+    # Overrides the `lvm_disk_paths` package-scoped fixture from the parent
+    # package so we can transparently use other fixtures that depend on it
+    # whilst having a dm-flakey device mapper underneath.
+    return {host: [disk.path] for (host, disk) in flakey_unused_512B_disk.items()}

tests/storage/linstor/unhealthy/conftest.py

@@ -0,0 +1,43 @@
+from __future__ import annotations
+
+import pytest
+
+import logging
+import random
+
+from lib.host import Host
+from lib.sr import SR
+from lib.vm import VM
+
+from .conftest import FlakeyDisk
+
+# Requirements:
+# - three or more XCP-ng hosts >= 8.2 with additional unused disk(s) for the SR
+# - access to XCP-ng RPM repository from the host
+
+class TestLinstorSRFailedDisk:
+    @pytest.mark.small_vm # run with a small VM to test the features
+    def test_linstor_sr_fail_disk(
+        self,
+        vm_on_linstor_sr: VM,
+        flakey_unused_512B_disk: dict[Host, FlakeyDisk],
+        linstor_sr: SR,
+    ) -> None:
+        sr = linstor_sr
+        vm = vm_on_linstor_sr
+        random_host = random.choice(sr.pool.hosts)
+
+        # Let xcp-persistent-database come in sync across the nodes.
+        random_host.ssh('drbdadm wait-sync xcp-persistent-database')
+
+        flakey_unused_512B_disk[random_host].fail()
+
+        try:
+            for host in sr.pool.hosts:
+                logging.info(f'Checking VM on host {host.hostname_or_ip}')
+
+                vm.start(on=host.uuid)
+                vm.wait_for_os_booted()
+                vm.shutdown(verify=True)
+        finally:
+            flakey_unused_512B_disk[random_host].repair()