import os import subprocess import time import infra.basetest class TestMdadm(infra.basetest.BRTest): # This test creates a dm-raid volume with mdadm. A specific Kernel # need to be built with a config fragment enabling this support. kernel_fragment = \ infra.filepath("tests/package/test_mdadm/linux-mdadm.fragment") config = \ f""" BR2_aarch64=y BR2_TOOLCHAIN_EXTERNAL=y BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0" BR2_LINUX_KERNEL=y BR2_LINUX_KERNEL_CUSTOM_VERSION=y BR2_LINUX_KERNEL_CUSTOM_VERSION_VALUE="6.1.75" BR2_LINUX_KERNEL_USE_CUSTOM_CONFIG=y BR2_LINUX_KERNEL_CUSTOM_CONFIG_FILE="board/qemu/aarch64-virt/linux.config" BR2_LINUX_KERNEL_CONFIG_FRAGMENT_FILES="{kernel_fragment}" BR2_LINUX_KERNEL_NEEDS_HOST_OPENSSL=y BR2_PACKAGE_E2FSPROGS=y BR2_PACKAGE_MDADM=y BR2_TARGET_ROOTFS_CPIO=y BR2_TARGET_ROOTFS_CPIO_GZIP=y # BR2_TARGET_ROOTFS_TAR is not set """ def test_run(self): # Test configuration: md_dev = "/dev/md0" storage_devs = ["/dev/vda", "/dev/vdb", "/dev/vdc"] storage_size = 16 # Mega Bytes failing_dev = storage_devs[-1] mnt_pt = "/mnt/raid-storage" data_file = f"{mnt_pt}/data.bin" qemu_storage_opts = [] for i in range(len(storage_devs)): disk_file = os.path.join(self.builddir, "images", f"disk{i}.img") self.emulator.logfile.write(f"Creating disk image: {disk_file}\n") self.emulator.logfile.flush() subprocess.check_call( ["dd", "if=/dev/zero", f"of={disk_file}", "bs=1M", f"count={storage_size}"], stdout=self.emulator.logfile, stderr=self.emulator.logfile) opts = ["-drive", f"file={disk_file},if=virtio,format=raw"] qemu_storage_opts += opts img = os.path.join(self.builddir, "images", "rootfs.cpio.gz") kern = os.path.join(self.builddir, "images", "Image") self.emulator.boot(arch="aarch64", kernel=kern, kernel_cmdline=["console=ttyAMA0"], options=["-M", "virt", "-cpu", "cortex-a57", "-m", "256M", "-initrd", img] + qemu_storage_opts) self.emulator.login() # Test the program can execute. self.assertRunOk("mdadm --version") # Show the mdstat, to confirm the Kernel has support and the # configuration is empty. cat_mdstat_cmd = "cat /proc/mdstat" self.assertRunOk(cat_mdstat_cmd) # We create a raid5 array with the drives. cmd = f"mdadm --create --verbose {md_dev} --level=5 " cmd += f"--raid-devices={len(storage_devs)} " cmd += " ".join(storage_devs) self.assertRunOk(cmd) # We show again mdstat, to confirm the array creation. This is # also for debugging. self.assertRunOk(cat_mdstat_cmd) # We format the device as ext4 and mount it. self.assertRunOk(f"mkfs.ext4 {md_dev}") self.assertRunOk(f"mkdir -p {mnt_pt}") self.assertRunOk(f"mount {md_dev} {mnt_pt}") # We store some random data on this new filesystem. Note: this # file is slightly larger than a single storage drive. This # data file should span over two drives and use the raid5. file_size = storage_size + 4 cmd = f"dd if=/dev/urandom of={data_file} bs=1M count={file_size}" self.assertRunOk(cmd) # We compute the hash of our data, and save it for later. hash_cmd = f"sha256sum {data_file}" out, ret = self.emulator.run(hash_cmd) self.assertEqual(ret, 0) data_sha256 = out[0] # We run few common mdadm commands. self.assertRunOk("mdadm --detail --scan") self.assertRunOk(f"mdadm --query {md_dev}") self.assertRunOk(f"mdadm --detail --test {md_dev}") self.assertRunOk(f"mdadm --action=check {md_dev}") self.assertRunOk(f"mdadm --monitor --oneshot {md_dev}") # We mark a device as "failed". self.assertRunOk(f"mdadm {md_dev} --fail {failing_dev}") # The monitoring should now report the array as degraded. monitor_cmd = f"mdadm --monitor --oneshot {md_dev}" out, ret = self.emulator.run(monitor_cmd) self.assertEqual(ret, 0) self.assertIn("DegradedArray", "\n".join(out)) # We remove the failing drive from the array. self.assertRunOk(f"mdadm {md_dev} --remove {failing_dev}") # We wipe the failing drive by writing zeroes. cmd = f"dd if=/dev/zero of={failing_dev} bs=1M count={storage_size}" self.assertRunOk(cmd) # We add back this blank drive to the array. self.assertRunOk(f"mdadm {md_dev} --add {failing_dev}") # Device rebuild can take a variable amount of time, depending # on the load of the test controller host. So we will allow # several attempts, before failing. for attempt in range(10): # We wait few seconds to let the device rebuild. time.sleep(3 * self.timeout_multiplier) # Once rebuilt, the array should no longer be marked as # degraded. out, ret = self.emulator.run(monitor_cmd) self.assertEqual(ret, 0) if "DegradedArray" not in "\n".join(out): break else: self.fail("Timeout while waiting for the array to rebuild.") # With all those array manipulations, the data file should not # be corrupted. We should be able to recompute the same hash # as before. out, ret = self.emulator.run(hash_cmd) self.assertEqual(ret, 0) self.assertEqual(out[0], data_sha256)