open-cas-linux/test/functional/tests/io_class/test_io_class_file.py
Rafal Stefanowski acec05060d Fix license
Change license to BSD-3-Clause

Signed-off-by: Rafal Stefanowski <rafal.stefanowski@intel.com>
2021-10-28 12:46:42 +02:00

523 lines
20 KiB
Python

#
# Copyright(c) 2019-2021 Intel Corporation
# SPDX-License-Identifier: BSD-3-Clause
#
import random
import pytest
from api.cas import ioclass_config, casadm
from core.test_run import TestRun
from storage_devices.disk import DiskType, DiskTypeSet, DiskTypeLowerThan
from test_tools.dd import Dd
from test_tools.disk_utils import Filesystem
from test_utils.filesystem.file import File
from test_utils.os_utils import sync, DropCachesMode, drop_caches
from test_utils.size import Size, Unit
from tests.io_class.io_class_common import mountpoint, prepare, ioclass_config_path
@pytest.mark.require_disk("cache", DiskTypeSet([DiskType.optane, DiskType.nand]))
@pytest.mark.require_disk("core", DiskTypeLowerThan("cache"))
def test_ioclass_file_extension():
"""
title: Test IO classification by file extension.
description: Test if file extension classification works properly.
pass_criteria:
- No kernel bug.
- IO is classified properly based on IO class rule with file extension.
"""
iterations = 50
ioclass_id = 1
tested_extension = "tmp"
wrong_extensions = ["tm", "tmpx", "txt", "t", "", "123", "tmp.xx"]
dd_size = Size(4, Unit.KibiByte)
dd_count = 10
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/test_file.{tested_extension}")
.count(dd_count)
.block_size(dd_size)
)
with TestRun.step("Prepare cache and core."):
cache, core = prepare()
with TestRun.step("Create and load IO class config."):
ioclass_config.add_ioclass(
ioclass_id=ioclass_id,
eviction_priority=1,
allocation="1.00",
rule=f"extension:{tested_extension}&done",
ioclass_config_path=ioclass_config_path,
)
casadm.load_io_classes(cache_id=cache.cache_id, file=ioclass_config_path)
with TestRun.step(f"Prepare filesystem and mount {core.path} at {mountpoint}."):
core.create_filesystem(Filesystem.ext3)
core.mount(mountpoint)
with TestRun.step("Flush cache."):
cache.flush_cache()
with TestRun.step(f"Write to file with cached extension and check if it is properly cached."):
for i in range(iterations):
dd.run()
sync()
dirty = cache.get_io_class_statistics(io_class_id=ioclass_id).usage_stats.dirty
if dirty.get_value(Unit.Blocks4096) != (i + 1) * dd_count:
TestRun.LOGGER.error(f"Wrong amount of dirty data ({dirty}).")
with TestRun.step("Flush cache."):
cache.flush_cache()
with TestRun.step(f"Write to file with not cached extension and check if it is not cached."):
for ext in wrong_extensions:
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/test_file.{ext}")
.count(dd_count)
.block_size(dd_size)
)
dd.run()
sync()
dirty = cache.get_io_class_statistics(io_class_id=ioclass_id).usage_stats.dirty
if dirty.get_value(Unit.Blocks4096) != 0:
TestRun.LOGGER.error(f"Wrong amount of dirty data ({dirty}).")
@pytest.mark.require_disk("cache", DiskTypeSet([DiskType.optane, DiskType.nand]))
@pytest.mark.require_disk("core", DiskTypeLowerThan("cache"))
def test_ioclass_file_name_prefix():
"""
title: Test IO classification by file name prefix.
description: Test if file name prefix classification works properly.
pass_criteria:
- No kernel bug.
- IO is classified properly based on IO class rule with file name prefix.
"""
ioclass_id = 1
cached_files = ["test", "test.txt", "test1", "test1.txt"]
not_cached_files = ["file1", "file2", "file4", "file5", "tes"]
dd_size = Size(4, Unit.KibiByte)
dd_count = 10
with TestRun.step("Prepare cache and core."):
cache, core = prepare()
with TestRun.step("Create and load IO class config."):
ioclass_config.remove_ioclass_config()
ioclass_config.create_ioclass_config(False)
# Avoid caching anything else than files with specified prefix
ioclass_config.add_ioclass(
ioclass_id=0,
eviction_priority=255,
allocation="0.00",
rule=f"unclassified",
ioclass_config_path=ioclass_config_path,
)
# Enables file with specified prefix to be cached
ioclass_config.add_ioclass(
ioclass_id=ioclass_id,
eviction_priority=1,
allocation="1.00",
rule=f"file_name_prefix:test&done",
ioclass_config_path=ioclass_config_path,
)
casadm.load_io_classes(cache_id=cache.cache_id, file=ioclass_config_path)
with TestRun.step(f"Prepare filesystem and mount {core.path} at {mountpoint}"):
previous_occupancy = cache.get_occupancy()
core.create_filesystem(Filesystem.ext3)
core.mount(mountpoint)
current_occupancy = cache.get_occupancy()
if previous_occupancy.get_value() > current_occupancy.get_value():
TestRun.fail(f"Current occupancy ({str(current_occupancy)}) is lower "
f"than before ({str(previous_occupancy)}).")
# Filesystem creation caused metadata IO which is not supposed
# to be cached
# Check if files with proper prefix are cached
with TestRun.step(f"Write files which are supposed to be cached and check "
f"if they are cached."):
for f in cached_files:
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/{f}")
.count(dd_count)
.block_size(dd_size)
)
dd.run()
sync()
current_occupancy = cache.get_occupancy()
expected_occupancy = previous_occupancy + (dd_size * dd_count)
if current_occupancy != expected_occupancy:
TestRun.fail(f"Current occupancy value is not valid. "
f"(Expected: {str(expected_occupancy)}, "
f"actual: {str(current_occupancy)})")
previous_occupancy = current_occupancy
with TestRun.step("Flush cache."):
cache.flush_cache()
# Check if file with improper extension is not cached
with TestRun.step(f"Write files which are not supposed to be cached and check if "
f"they are not cached."):
for f in not_cached_files:
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/{f}")
.count(dd_count)
.block_size(dd_size)
)
dd.run()
sync()
current_occupancy = cache.get_occupancy()
if current_occupancy != previous_occupancy:
TestRun.fail(f"Current occupancy value is not valid. "
f"(Expected: {str(previous_occupancy)}, "
f"actual: {str(current_occupancy)})")
@pytest.mark.require_disk("cache", DiskTypeSet([DiskType.optane, DiskType.nand]))
@pytest.mark.require_disk("core", DiskTypeLowerThan("cache"))
def test_ioclass_file_extension_preexisting_filesystem():
"""
title: Test IO classification by file extension with preexisting filesystem on core device.
description: |
Test if file extension classification works properly when there is an existing
filesystem on core device.
pass_criteria:
- No kernel bug.
- IO is classified properly based on IO class rule with file extension
after mounting core device.
"""
ioclass_id = 1
extensions = ["tmp", "tm", "out", "txt", "log", "123"]
dd_size = Size(4, Unit.KibiByte)
dd_count = 10
with TestRun.step("Prepare cache and core devices."):
cache, core = prepare()
with TestRun.step(f"Prepare files on raw block device."):
casadm.remove_core(cache.cache_id, core_id=core.core_id)
core.core_device.create_filesystem(Filesystem.ext3)
core.core_device.mount(mountpoint)
for ext in extensions:
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/test_file.{ext}")
.count(dd_count)
.block_size(dd_size)
)
dd.run()
core.core_device.unmount()
with TestRun.step("Create IO class config."):
rule = "|".join([f"extension:{ext}" for ext in extensions])
ioclass_config.add_ioclass(
ioclass_id=ioclass_id,
eviction_priority=1,
allocation="1.00",
rule=f"{rule}&done",
ioclass_config_path=ioclass_config_path,
)
with TestRun.step(f"Add device with preexisting data as a core."):
core = casadm.add_core(cache, core_dev=core.core_device)
with TestRun.step("Load IO class config."):
casadm.load_io_classes(cache_id=cache.cache_id, file=ioclass_config_path)
with TestRun.step("Mount core and flush cache."):
core.mount(mountpoint)
cache.flush_cache()
with TestRun.step(f"Write to file with cached extension and check if they are cached."):
for ext in extensions:
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/test_file.{ext}")
.count(dd_count)
.block_size(dd_size)
)
dd.run()
sync()
dirty = cache.get_io_class_statistics(io_class_id=ioclass_id).usage_stats.dirty
if dirty.get_value(Unit.Blocks4096) != (extensions.index(ext) + 1) * dd_count:
TestRun.LOGGER.error(f"Wrong amount of dirty data ({dirty}).")
@pytest.mark.require_disk("cache", DiskTypeSet([DiskType.optane, DiskType.nand]))
@pytest.mark.require_disk("core", DiskTypeLowerThan("cache"))
def test_ioclass_file_offset():
"""
title: Test IO classification by file offset.
description: Test if file offset classification works properly.
pass_criteria:
- No kernel bug.
- IO is classified properly based on IO class rule with file offset.
"""
ioclass_id = 1
iterations = 100
dd_size = Size(4, Unit.KibiByte)
dd_count = 1
min_cached_offset = 16384
max_cached_offset = 65536
with TestRun.step("Prepare cache and core."):
cache, core = prepare()
with TestRun.step("Create and load IO class config file."):
ioclass_config.add_ioclass(
ioclass_id=ioclass_id,
eviction_priority=1,
allocation="1.00",
rule=f"file_offset:gt:{min_cached_offset}&file_offset:lt:{max_cached_offset}&done",
ioclass_config_path=ioclass_config_path,
)
casadm.load_io_classes(cache_id=cache.cache_id, file=ioclass_config_path)
with TestRun.step(f"Prepare filesystem and mount {core.path} at {mountpoint}."):
core.create_filesystem(Filesystem.ext3)
core.mount(mountpoint)
with TestRun.step("Flush cache."):
cache.flush_cache()
with TestRun.step("Write to file within cached offset range and check if it is cached."):
# Since ioclass rule consists of strict inequalities, 'seek' can't be set to first
# nor last sector
min_seek = int((min_cached_offset + Unit.Blocks4096.value) / Unit.Blocks4096.value)
max_seek = int(
(max_cached_offset - min_cached_offset - Unit.Blocks4096.value)
/ Unit.Blocks4096.value
)
for i in range(iterations):
file_offset = random.choice(range(min_seek, max_seek))
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/tmp_file")
.count(dd_count)
.block_size(dd_size)
.seek(file_offset)
)
dd.run()
sync()
dirty = cache.get_io_class_statistics(io_class_id=ioclass_id).usage_stats.dirty
if dirty.get_value(Unit.Blocks4096) != 1:
TestRun.LOGGER.error(f"Offset not cached: {file_offset}")
cache.flush_cache()
with TestRun.step(
"Write to file outside of cached offset range and check if it is not cached."):
min_seek = 0
max_seek = int(min_cached_offset / Unit.Blocks4096.value)
TestRun.LOGGER.info(f"Writing to file outside of cached offset range")
for i in range(iterations):
file_offset = random.choice(range(min_seek, max_seek))
dd = (
Dd().input("/dev/zero")
.output(f"{mountpoint}/tmp_file")
.count(dd_count)
.block_size(dd_size)
.seek(file_offset)
)
dd.run()
sync()
dirty = cache.get_io_class_statistics(io_class_id=ioclass_id).usage_stats.dirty
if dirty.get_value(Unit.Blocks4096) != 0:
TestRun.LOGGER.error(f"Inappropriately cached offset: {file_offset}")
@pytest.mark.os_dependent
@pytest.mark.require_disk("cache", DiskTypeSet([DiskType.optane, DiskType.nand]))
@pytest.mark.require_disk("core", DiskTypeLowerThan("cache"))
@pytest.mark.parametrizex("filesystem", Filesystem)
def test_ioclass_file_size(filesystem):
"""
title: Test IO classification by file size.
description: Test if file size classification works properly.
pass_criteria:
- No kernel bug.
- IO is classified properly based on IO class rule with file size.
"""
# File size IO class rules are configured in a way that each tested file size is unambiguously
# classified.
# Firstly write operations are tested (creation of files), secondly read operations.
base_size = Size(random.randint(50, 1000) * 2, Unit.Blocks4096)
size_to_class = {
base_size: 1,
base_size - Unit.Blocks4096: 2,
base_size + Unit.Blocks4096: 3,
base_size / 2: 4,
base_size / 2 - Unit.Blocks4096: 4,
base_size / 2 + Unit.Blocks4096: 2,
base_size * 2: 5,
base_size * 2 - Unit.Blocks4096: 3,
base_size * 2 + Unit.Blocks4096: 5,
}
with TestRun.step("Prepare cache and core."):
cache, core = prepare(default_allocation="1.00")
with TestRun.step("Prepare and load IO class config."):
load_file_size_io_classes(cache, base_size)
with TestRun.step(f"Prepare {filesystem.name} filesystem and mount {core.path} "
f"at {mountpoint}."):
core.create_filesystem(filesystem)
core.mount(mountpoint)
sync()
with TestRun.step("Create files belonging to different IO classes (classification by writes)."):
test_files = []
for size, ioclass_id in size_to_class.items():
occupancy_before = cache.get_io_class_statistics(
io_class_id=ioclass_id).usage_stats.occupancy
file_path = f"{mountpoint}/test_file_{size.get_value()}"
Dd().input("/dev/zero").output(file_path).oflag("sync").block_size(size).count(1).run()
sync()
drop_caches(DropCachesMode.ALL)
occupancy_after = cache.get_io_class_statistics(
io_class_id=ioclass_id).usage_stats.occupancy
if occupancy_after != occupancy_before + size:
TestRun.fail("File not cached properly!\n"
f"Expected {occupancy_before + size}\n"
f"Actual {occupancy_after}")
test_files.append(File(file_path).refresh_item())
sync()
drop_caches(DropCachesMode.ALL)
with TestRun.step("Move all files to 'unclassified' IO class."):
ioclass_config.remove_ioclass_config(ioclass_config_path=ioclass_config_path)
ioclass_config.create_ioclass_config(
add_default_rule=False, ioclass_config_path=ioclass_config_path
)
ioclass_config.add_ioclass(
ioclass_id=0,
eviction_priority=22,
allocation="1.00",
rule="unclassified",
ioclass_config_path=ioclass_config_path,
)
ioclass_config.add_ioclass(
ioclass_id=6,
eviction_priority=1,
allocation="0.00",
rule=f"metadata",
ioclass_config_path=ioclass_config_path,
)
casadm.load_io_classes(cache_id=cache.cache_id, file=ioclass_config_path)
occupancy_before = cache.get_io_class_statistics(io_class_id=0).usage_stats.occupancy
for file in test_files:
Dd().input(file.full_path).output("/dev/null").block_size(file.size).run()
sync()
drop_caches(DropCachesMode.ALL)
occupancy_after = cache.get_io_class_statistics(io_class_id=0).usage_stats.occupancy
occupancy_expected = occupancy_before + file.size
if occupancy_after != occupancy_expected:
TestRun.fail("File not reclassified properly!\n"
f"Expected {occupancy_expected}\n"
f"Actual {occupancy_after}")
occupancy_before = occupancy_after
sync()
drop_caches(DropCachesMode.ALL)
with TestRun.step("Restore IO class configuration."):
ioclass_config.remove_ioclass_config(ioclass_config_path=ioclass_config_path)
ioclass_config.create_ioclass_config(
add_default_rule=False, ioclass_config_path=ioclass_config_path
)
ioclass_config.add_ioclass(
ioclass_id=0,
eviction_priority=22,
allocation="1.00",
rule="unclassified",
ioclass_config_path=ioclass_config_path,
)
load_file_size_io_classes(cache, base_size)
with TestRun.step("Read files belonging to different IO classes (classification by reads)."):
# CAS device should be unmounted and mounted because data can be sometimes still cached by
# OS cache so occupancy statistics will not match
core.unmount()
core.mount(mountpoint)
for file in test_files:
ioclass_id = size_to_class[file.size]
occupancy_before = cache.get_io_class_statistics(
io_class_id=ioclass_id).usage_stats.occupancy
Dd().input(file.full_path).output("/dev/null").block_size(file.size).run()
sync()
drop_caches(DropCachesMode.ALL)
occupancy_after = cache.get_io_class_statistics(
io_class_id=ioclass_id).usage_stats.occupancy
actual_blocks = occupancy_after.get_value(Unit.Blocks4096)
expected_blocks = (occupancy_before + file.size).get_value(Unit.Blocks4096)
if actual_blocks != expected_blocks:
TestRun.fail("File not reclassified properly!\n"
f"Expected {occupancy_before + file.size}\n"
f"Actual {occupancy_after}")
sync()
drop_caches(DropCachesMode.ALL)
def load_file_size_io_classes(cache, base_size):
# IO class order intentional, do not change
base_size_bytes = int(base_size.get_value(Unit.Byte))
ioclass_config.add_ioclass(
ioclass_id=6,
eviction_priority=1,
allocation="0.00",
rule=f"metadata",
ioclass_config_path=ioclass_config_path,
)
ioclass_config.add_ioclass(
ioclass_id=1,
eviction_priority=1,
allocation="1.00",
rule=f"file_size:eq:{base_size_bytes}",
ioclass_config_path=ioclass_config_path,
)
ioclass_config.add_ioclass(
ioclass_id=2,
eviction_priority=1,
allocation="1.00",
rule=f"file_size:lt:{base_size_bytes}",
ioclass_config_path=ioclass_config_path,
)
ioclass_config.add_ioclass(
ioclass_id=3,
eviction_priority=1,
allocation="1.00",
rule=f"file_size:gt:{base_size_bytes}",
ioclass_config_path=ioclass_config_path,
)
ioclass_config.add_ioclass(
ioclass_id=4,
eviction_priority=1,
allocation="1.00",
rule=f"file_size:le:{int(base_size_bytes / 2)}",
ioclass_config_path=ioclass_config_path,
)
ioclass_config.add_ioclass(
ioclass_id=5,
eviction_priority=1,
allocation="1.00",
rule=f"file_size:ge:{2 * base_size_bytes}",
ioclass_config_path=ioclass_config_path,
)
casadm.load_io_classes(cache_id=cache.cache_id, file=ioclass_config_path)