You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
rocksdb/tools/db_crashtest.py

730 lines
29 KiB

#!/usr/bin/env python3
# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
from __future__ import absolute_import, division, print_function, unicode_literals
import os
import sys
import time
import random
import re
import tempfile
import subprocess
import shutil
import argparse
# params overwrite priority:
# for default:
# default_params < {blackbox,whitebox}_default_params < args
# for simple:
# default_params < {blackbox,whitebox}_default_params <
# simple_default_params <
# {blackbox,whitebox}_simple_default_params < args
# for cf_consistency:
# default_params < {blackbox,whitebox}_default_params <
# cf_consistency_params < args
# for txn:
# default_params < {blackbox,whitebox}_default_params < txn_params < args
default_params = {
"acquire_snapshot_one_in": 10000,
"backup_max_size": 100 * 1024 * 1024,
# Consider larger number when backups considered more stable
"backup_one_in": 100000,
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
4 years ago
"batch_protection_bytes_per_key": lambda: random.choice([0, 8]),
"block_size": 16384,
"bloom_bits": lambda: random.choice([random.randint(0,19),
random.lognormvariate(2.3, 1.3)]),
"cache_index_and_filter_blocks": lambda: random.randint(0, 1),
"cache_size": 1048576,
"checkpoint_one_in": 1000000,
"compression_type": lambda: random.choice(
["none", "snappy", "zlib", "bzip2", "lz4", "lz4hc", "xpress", "zstd"]),
"bottommost_compression_type": lambda:
"disable" if random.randint(0, 1) == 0 else
random.choice(
["none", "snappy", "zlib", "bzip2", "lz4", "lz4hc", "xpress",
"zstd"]),
"checksum_type" : lambda: random.choice(["kCRC32c", "kxxHash", "kxxHash64"]),
"compression_max_dict_bytes": lambda: 16384 * random.randint(0, 1),
"compression_zstd_max_train_bytes": lambda: 65536 * random.randint(0, 1),
# Disabled compression_parallel_threads as the feature is not stable
# lambda: random.choice([1] * 9 + [4])
"compression_parallel_threads": 1,
Limit buffering for collecting samples for compression dictionary (#7970) Summary: For dictionary compression, we need to collect some representative samples of the data to be compressed, which we use to either generate or train (when `CompressionOptions::zstd_max_train_bytes > 0`) a dictionary. Previously, the strategy was to buffer all the data blocks during flush, and up to the target file size during compaction. That strategy allowed us to randomly pick samples from as wide a range as possible that'd be guaranteed to land in a single output file. However, some users try to make huge files in memory-constrained environments, where this strategy can cause OOM. This PR introduces an option, `CompressionOptions::max_dict_buffer_bytes`, that limits how much data blocks are buffered before we switch to unbuffered mode (which means creating the per-SST dictionary, writing out the buffered data, and compressing/writing new blocks as soon as they are built). It is not strict as we currently buffer more than just data blocks -- also keys are buffered. But it does make a step towards giving users predictable memory usage. Related changes include: - Changed sampling for dictionary compression to select unique data blocks when there is limited availability of data blocks - Made use of `BlockBuilder::SwapAndReset()` to save an allocation+memcpy when buffering data blocks for building a dictionary - Changed `ParseBoolean()` to accept an input containing characters after the boolean. This is necessary since, with this PR, a value for `CompressionOptions::enabled` is no longer necessarily the final component in the `CompressionOptions` string. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7970 Test Plan: - updated `CompressionOptions` unit tests to verify limit is respected (to the extent expected in the current implementation) in various scenarios of flush/compaction to bottommost/non-bottommost level - looked at jemalloc heap profiles right before and after switching to unbuffered mode during flush/compaction. Verified memory usage in buffering is proportional to the limit set. Reviewed By: pdillinger Differential Revision: D26467994 Pulled By: ajkr fbshipit-source-id: 3da4ef9fba59974e4ef40e40c01611002c861465
4 years ago
"compression_max_dict_buffer_bytes": lambda: (1 << random.randint(0, 40)) - 1,
"clear_column_family_one_in": 0,
"compact_files_one_in": 1000000,
"compact_range_one_in": 1000000,
"delpercent": 4,
"delrangepercent": 1,
"destroy_db_initially": 0,
"enable_pipelined_write": lambda: random.randint(0, 1),
"enable_compaction_filter": lambda: random.choice([0, 0, 0, 1]),
Refactor expected state in stress/crash test (#8913) Summary: This is a precursor refactoring to enable an upcoming feature: persistence failure correctness testing. - Changed `--expected_values_path` to `--expected_values_dir` and migrated "db_crashtest.py" to use the new flag. For persistence failure correctness testing there are multiple possible correct states since unsynced data is allowed to be dropped. Making it possible to restore all these possible correct states will eventually involve files containing snapshots of expected values and DB trace files. - The expected values directory is managed by an `ExpectedStateManager` instance. Managing expected state files is separated out of `SharedState` to prevent `SharedState` from becoming too complex when the new files and features (snapshotting, tracing, and restoring) are introduced. - Migrated expected values file access/management out of `SharedState` into a separate class called `ExpectedState`. This is not exposed directly to the test but rather the `ExpectedState` for the latest values file is accessed via a pass-through API on `ExpectedStateManager`. This forces the test to always access the single latest `ExpectedState`. - Changed the initialization of the latest expected values file to use a tempfile followed by rename, and also add cleanup logic for possible stranded tempfiles. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8913 Test Plan: run in several ways; try to make sure it's not obviously broken. - crashtest blackbox without TEST_TMPDIR ``` $ python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest blackbox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest whitebox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py whitebox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none --random_kill_odd=88887 ``` - db_stress without expected_values_dir ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true ``` - db_stress with expected_values_dir and manual corruption ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true --expected_values_dir=./ // modify one byte in "./LATEST.state" $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=false --expected_values_dir=./ ... Verification failed for column family 0 key 0000000000000000 (0): Value not found: NotFound: ... ``` Reviewed By: riversand963 Differential Revision: D30921951 Pulled By: ajkr fbshipit-source-id: babfe218062e55d018c9b046536c0289fb78f41c
3 years ago
"expected_values_dir": lambda: setup_expected_values_dir(),
"fail_if_options_file_error": lambda: random.randint(0, 1),
"flush_one_in": 1000000,
"file_checksum_impl": lambda: random.choice(["none", "crc32c", "xxh64", "big"]),
"get_live_files_one_in": 1000000,
# Note: the following two are intentionally disabled as the corresponding
# APIs are not guaranteed to succeed.
"get_sorted_wal_files_one_in": 0,
"get_current_wal_file_one_in": 0,
# Temporarily disable hash index
"index_type": lambda: random.choice([0, 0, 0, 2, 2, 3]),
"iterpercent": 10,
"mark_for_compaction_one_file_in": lambda: 10 * random.randint(0, 1),
"max_background_compactions": 20,
"max_bytes_for_level_base": 10485760,
"max_key": 100000000,
"max_write_buffer_number": 3,
"mmap_read": lambda: random.randint(0, 1),
Protect existing files in `FaultInjectionTest{Env,FS}::ReopenWritableFile()` (#8995) Summary: `FaultInjectionTest{Env,FS}::ReopenWritableFile()` functions were accidentally deleting WALs from previous `db_stress` runs causing verification to fail. They were operating under the assumption that `ReopenWritableFile()` would delete any existing file. It was a reasonable assumption considering the `{Env,FileSystem}::ReopenWritableFile()` documentation stated that would happen. The only problem was neither the implementations we offer nor the "real" clients in RocksDB code followed that contract. So, this PR updates the contract as well as fixing the fault injection client usage. The fault injection change exposed that `ExternalSSTFileBasicTest.SyncFailure` was relying on a fault injection `Env` dropping unsynced data written by a regular `Env`. I changed that test to make its `SstFileWriter` use fault injection `Env`, and also implemented `LinkFile()` in fault injection so the unsynced data is tracked under the new name. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8995 Test Plan: - Verified it fixes the following failure: ``` $ ./db_stress --clear_column_family_one_in=0 --column_families=1 --db=/dev/shm/rocksdb_crashtest_whitebox --delpercent=5 --expected_values_dir=/dev/shm/rocksdb_crashtest_expected --iterpercent=0 --key_len_percent_dist=1,30,69 --max_key=100000 --max_key_len=3 --nooverwritepercent=1 --ops_per_thread=1000 --prefixpercent=0 --readpercent=60 --reopen=0 --target_file_size_base=1048576 --test_batches_snapshots=0 --write_buffer_size=1048576 --writepercent=35 --value_size_mult=33 -threads=1 ... $ ./db_stress --avoid_flush_during_recovery=1 --clear_column_family_one_in=0 --column_families=1 --db=/dev/shm/rocksdb_crashtest_whitebox --delpercent=5 --destroy_db_initially=0 --expected_values_dir=/dev/shm/rocksdb_crashtest_expected --iterpercent=10 --key_len_percent_dist=1,30,69 --max_bytes_for_level_base=4194304 --max_key=100000 --max_key_len=3 --nooverwritepercent=1 --open_files=-1 --open_metadata_write_fault_one_in=8 --open_write_fault_one_in=16 --ops_per_thread=1000 --prefix_size=-1 --prefixpercent=0 --readpercent=50 --sync=1 --target_file_size_base=1048576 --test_batches_snapshots=0 --write_buffer_size=1048576 --writepercent=35 --value_size_mult=33 -threads=1 ... Verification failed for column family 0 key 000000000000001300000000000000857878787878 (1143): Value not found: NotFound: Crash-recovery verification failed :( ... ``` - `make check -j48` Reviewed By: ltamasi Differential Revision: D31495388 Pulled By: ajkr fbshipit-source-id: 7886ccb6a07cb8b78ad7b6c1c341ccf40bb68385
3 years ago
# Setting `nooverwritepercent > 0` is only possible because we do not vary
# the random seed, so the same keys are chosen by every run for disallowing
# overwrites.
"nooverwritepercent": 1,
"open_files": lambda : random.choice([-1, -1, 100, 500000]),
Minimize memory internal fragmentation for Bloom filters (#6427) Summary: New experimental option BBTO::optimize_filters_for_memory builds filters that maximize their use of "usable size" from malloc_usable_size, which is also used to compute block cache charges. Rather than always "rounding up," we track state in the BloomFilterPolicy object to mix essentially "rounding down" and "rounding up" so that the average FP rate of all generated filters is the same as without the option. (YMMV as heavily accessed filters might be unluckily lower accuracy.) Thus, the option near-minimizes what the block cache considers as "memory used" for a given target Bloom filter false positive rate and Bloom filter implementation. There are no forward or backward compatibility issues with this change, though it only works on the format_version=5 Bloom filter. With Jemalloc, we see about 10% reduction in memory footprint (and block cache charge) for Bloom filters, but 1-2% increase in storage footprint, due to encoding efficiency losses (FP rate is non-linear with bits/key). Why not weighted random round up/down rather than state tracking? By only requiring malloc_usable_size, we don't actually know what the next larger and next smaller usable sizes for the allocator are. We pick a requested size, accept and use whatever usable size it has, and use the difference to inform our next choice. This allows us to narrow in on the right balance without tracking/predicting usable sizes. Why not weight history of generated filter false positive rates by number of keys? This could lead to excess skew in small filters after generating a large filter. Results from filter_bench with jemalloc (irrelevant details omitted): (normal keys/filter, but high variance) $ ./filter_bench -quick -impl=2 -average_keys_per_filter=30000 -vary_key_count_ratio=0.9 Build avg ns/key: 29.6278 Number of filters: 5516 Total size (MB): 200.046 Reported total allocated memory (MB): 220.597 Reported internal fragmentation: 10.2732% Bits/key stored: 10.0097 Average FP rate %: 0.965228 $ ./filter_bench -quick -impl=2 -average_keys_per_filter=30000 -vary_key_count_ratio=0.9 -optimize_filters_for_memory Build avg ns/key: 30.5104 Number of filters: 5464 Total size (MB): 200.015 Reported total allocated memory (MB): 200.322 Reported internal fragmentation: 0.153709% Bits/key stored: 10.1011 Average FP rate %: 0.966313 (very few keys / filter, optimization not as effective due to ~59 byte internal fragmentation in blocked Bloom filter representation) $ ./filter_bench -quick -impl=2 -average_keys_per_filter=1000 -vary_key_count_ratio=0.9 Build avg ns/key: 29.5649 Number of filters: 162950 Total size (MB): 200.001 Reported total allocated memory (MB): 224.624 Reported internal fragmentation: 12.3117% Bits/key stored: 10.2951 Average FP rate %: 0.821534 $ ./filter_bench -quick -impl=2 -average_keys_per_filter=1000 -vary_key_count_ratio=0.9 -optimize_filters_for_memory Build avg ns/key: 31.8057 Number of filters: 159849 Total size (MB): 200 Reported total allocated memory (MB): 208.846 Reported internal fragmentation: 4.42297% Bits/key stored: 10.4948 Average FP rate %: 0.811006 (high keys/filter) $ ./filter_bench -quick -impl=2 -average_keys_per_filter=1000000 -vary_key_count_ratio=0.9 Build avg ns/key: 29.7017 Number of filters: 164 Total size (MB): 200.352 Reported total allocated memory (MB): 221.5 Reported internal fragmentation: 10.5552% Bits/key stored: 10.0003 Average FP rate %: 0.969358 $ ./filter_bench -quick -impl=2 -average_keys_per_filter=1000000 -vary_key_count_ratio=0.9 -optimize_filters_for_memory Build avg ns/key: 30.7131 Number of filters: 160 Total size (MB): 200.928 Reported total allocated memory (MB): 200.938 Reported internal fragmentation: 0.00448054% Bits/key stored: 10.1852 Average FP rate %: 0.963387 And from db_bench (block cache) with jemalloc: $ ./db_bench -db=/dev/shm/dbbench.no_optimize -benchmarks=fillrandom -format_version=5 -value_size=90 -bloom_bits=10 -num=2000000 -threads=8 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=false $ ./db_bench -db=/dev/shm/dbbench -benchmarks=fillrandom -format_version=5 -value_size=90 -bloom_bits=10 -num=2000000 -threads=8 -optimize_filters_for_memory -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=false $ (for FILE in /dev/shm/dbbench.no_optimize/*.sst; do ./sst_dump --file=$FILE --show_properties | grep 'filter block' ; done) | awk '{ t += $4; } END { print t; }' 17063835 $ (for FILE in /dev/shm/dbbench/*.sst; do ./sst_dump --file=$FILE --show_properties | grep 'filter block' ; done) | awk '{ t += $4; } END { print t; }' 17430747 $ #^ 2.1% additional filter storage $ ./db_bench -db=/dev/shm/dbbench.no_optimize -use_existing_db -benchmarks=readrandom,stats -statistics -bloom_bits=10 -num=2000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=false -duration=10 -cache_index_and_filter_blocks -cache_size=1000000000 rocksdb.block.cache.index.add COUNT : 33 rocksdb.block.cache.index.bytes.insert COUNT : 8440400 rocksdb.block.cache.filter.add COUNT : 33 rocksdb.block.cache.filter.bytes.insert COUNT : 21087528 rocksdb.bloom.filter.useful COUNT : 4963889 rocksdb.bloom.filter.full.positive COUNT : 1214081 rocksdb.bloom.filter.full.true.positive COUNT : 1161999 $ #^ 1.04 % observed FP rate $ ./db_bench -db=/dev/shm/dbbench -use_existing_db -benchmarks=readrandom,stats -statistics -bloom_bits=10 -num=2000000 -compaction_style=2 -fifo_compaction_max_table_files_size_mb=10000 -fifo_compaction_allow_compaction=false -optimize_filters_for_memory -duration=10 -cache_index_and_filter_blocks -cache_size=1000000000 rocksdb.block.cache.index.add COUNT : 33 rocksdb.block.cache.index.bytes.insert COUNT : 8448592 rocksdb.block.cache.filter.add COUNT : 33 rocksdb.block.cache.filter.bytes.insert COUNT : 18220328 rocksdb.bloom.filter.useful COUNT : 5360933 rocksdb.bloom.filter.full.positive COUNT : 1321315 rocksdb.bloom.filter.full.true.positive COUNT : 1262999 $ #^ 1.08 % observed FP rate, 13.6% less memory usage for filters (Due to specific key density, this example tends to generate filters that are "worse than average" for internal fragmentation. "Better than average" cases can show little or no improvement.) Pull Request resolved: https://github.com/facebook/rocksdb/pull/6427 Test Plan: unit test added, 'make check' with gcc, clang and valgrind Reviewed By: siying Differential Revision: D22124374 Pulled By: pdillinger fbshipit-source-id: f3e3aa152f9043ddf4fae25799e76341d0d8714e
4 years ago
"optimize_filters_for_memory": lambda: random.randint(0, 1),
"partition_filters": lambda: random.randint(0, 1),
"partition_pinning": lambda: random.randint(0, 3),
"pause_background_one_in": 1000000,
"prefixpercent": 5,
"progress_reports": 0,
"readpercent": 45,
"recycle_log_file_num": lambda: random.randint(0, 1),
"reopen": 20,
"snapshot_hold_ops": 100000,
"sst_file_manager_bytes_per_sec": lambda: random.choice([0, 104857600]),
"sst_file_manager_bytes_per_truncate": lambda: random.choice([0, 1048576]),
"long_running_snapshots": lambda: random.randint(0, 1),
"subcompactions": lambda: random.randint(1, 4),
"target_file_size_base": 2097152,
"target_file_size_multiplier": 2,
"top_level_index_pinning": lambda: random.randint(0, 3),
"unpartitioned_pinning": lambda: random.randint(0, 3),
"use_direct_reads": lambda: random.randint(0, 1),
"use_direct_io_for_flush_and_compaction": lambda: random.randint(0, 1),
"mock_direct_io": False,
"use_clock_cache": 0, # currently broken
"use_full_merge_v1": lambda: random.randint(0, 1),
"use_merge": lambda: random.randint(0, 1),
Add Bloom/Ribbon hybrid API support (#8679) Summary: This is essentially resurrection and fixing of the part of https://github.com/facebook/rocksdb/issues/8198 that was reverted in https://github.com/facebook/rocksdb/issues/8212, using data added in https://github.com/facebook/rocksdb/issues/8246. Basically, when configuring Ribbon filter, you can specify an LSM level before which Bloom will be used instead of Ribbon. But Bloom is only considered for Leveled and Universal compaction styles and file going into a known LSM level. This way, SST file writer, FIFO compaction, etc. use Ribbon filter as you would expect with NewRibbonFilterPolicy. So that this can be controlled with a single int value and so that flushes can be distinguished from intra-L0, we consider flush to go to level -1 for the purposes of this option. (Explained in API comment.) I also expect the most common and recommended Ribbon configuration to use Bloom during flush, to minimize slowing down writes and because according to my estimates, Ribbon only pays off if the structure lives in memory for more than an hour. Thus, I have changed the default for NewRibbonFilterPolicy to be this mild hybrid configuration. I don't really want to add something like NewHybridFilterPolicy because at least the mild hybrid configuration (Bloom for flush, Ribbon otherwise) should be considered a natural choice. C APIs also updated, but because they don't support overloading, rocksdb_filterpolicy_create_ribbon is kept pure ribbon for clarity and rocksdb_filterpolicy_create_ribbon_hybrid must be called for a hybrid configuration. While touching C API, I changed bits per key options from int to double. BuiltinFilterPolicy is needed so that LevelThresholdFilterPolicy doesn't inherit unused fields from BloomFilterPolicy. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8679 Test Plan: new + updated tests, including crash test Reviewed By: jay-zhuang Differential Revision: D30445797 Pulled By: pdillinger fbshipit-source-id: 6f5aeddfd6d79f7e55493b563c2d1d2d568892e1
3 years ago
# 999 -> use Bloom API
"ribbon_starting_level": lambda: random.choice([random.randint(-1, 10), 999]),
"use_block_based_filter": lambda: random.randint(0, 1),
"verify_checksum": 1,
"write_buffer_size": 4 * 1024 * 1024,
"writepercent": 35,
"format_version": lambda: random.choice([2, 3, 4, 5, 5]),
"index_block_restart_interval": lambda: random.choice(range(1, 16)),
"use_multiget" : lambda: random.randint(0, 1),
"periodic_compaction_seconds" :
lambda: random.choice([0, 0, 1, 2, 10, 100, 1000]),
"compaction_ttl" : lambda: random.choice([0, 0, 1, 2, 10, 100, 1000]),
# Test small max_manifest_file_size in a smaller chance, as most of the
# time we wnat manifest history to be preserved to help debug
"max_manifest_file_size" : lambda : random.choice(
[t * 16384 if t < 3 else 1024 * 1024 * 1024 for t in range(1, 30)]),
# Sync mode might make test runs slower so running it in a smaller chance
"sync" : lambda : random.choice(
[1 if t == 0 else 0 for t in range(0, 20)]),
# Disable compation_readahead_size because the test is not passing.
#"compaction_readahead_size" : lambda : random.choice(
# [0, 0, 1024 * 1024]),
"db_write_buffer_size" : lambda: random.choice(
[0, 0, 0, 1024 * 1024, 8 * 1024 * 1024, 128 * 1024 * 1024]),
"avoid_unnecessary_blocking_io" : random.randint(0, 1),
"write_dbid_to_manifest" : random.randint(0, 1),
"avoid_flush_during_recovery" : random.choice(
[1 if t == 0 else 0 for t in range(0, 8)]),
"max_write_batch_group_size_bytes" : lambda: random.choice(
[16, 64, 1024 * 1024, 16 * 1024 * 1024]),
"level_compaction_dynamic_level_bytes" : True,
"verify_checksum_one_in": 1000000,
"verify_db_one_in": 100000,
"continuous_verification_interval" : 0,
"max_key_len": 3,
"key_len_percent_dist": "1,30,69",
"read_fault_one_in": lambda: random.choice([0, 32, 1000]),
"open_metadata_write_fault_one_in": lambda: random.choice([0, 0, 8]),
"open_write_fault_one_in": lambda: random.choice([0, 0, 16]),
"open_read_fault_one_in": lambda: random.choice([0, 0, 32]),
"sync_fault_injection": False,
"get_property_one_in": 1000000,
"paranoid_file_checks": lambda: random.choice([0, 1, 1, 1]),
"max_write_buffer_size_to_maintain": lambda: random.choice(
[0, 1024 * 1024, 2 * 1024 * 1024, 4 * 1024 * 1024, 8 * 1024 * 1024]),
"user_timestamp_size": 0,
}
_TEST_DIR_ENV_VAR = 'TEST_TMPDIR'
_DEBUG_LEVEL_ENV_VAR = 'DEBUG_LEVEL'
stress_cmd = "./db_stress"
def is_release_mode():
return os.environ.get(_DEBUG_LEVEL_ENV_VAR) == "0"
def get_dbname(test_name):
test_dir_name = "rocksdb_crashtest_" + test_name
test_tmpdir = os.environ.get(_TEST_DIR_ENV_VAR)
if test_tmpdir is None or test_tmpdir == "":
dbname = tempfile.mkdtemp(prefix=test_dir_name)
else:
dbname = test_tmpdir + "/" + test_dir_name
shutil.rmtree(dbname, True)
os.mkdir(dbname)
return dbname
Refactor expected state in stress/crash test (#8913) Summary: This is a precursor refactoring to enable an upcoming feature: persistence failure correctness testing. - Changed `--expected_values_path` to `--expected_values_dir` and migrated "db_crashtest.py" to use the new flag. For persistence failure correctness testing there are multiple possible correct states since unsynced data is allowed to be dropped. Making it possible to restore all these possible correct states will eventually involve files containing snapshots of expected values and DB trace files. - The expected values directory is managed by an `ExpectedStateManager` instance. Managing expected state files is separated out of `SharedState` to prevent `SharedState` from becoming too complex when the new files and features (snapshotting, tracing, and restoring) are introduced. - Migrated expected values file access/management out of `SharedState` into a separate class called `ExpectedState`. This is not exposed directly to the test but rather the `ExpectedState` for the latest values file is accessed via a pass-through API on `ExpectedStateManager`. This forces the test to always access the single latest `ExpectedState`. - Changed the initialization of the latest expected values file to use a tempfile followed by rename, and also add cleanup logic for possible stranded tempfiles. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8913 Test Plan: run in several ways; try to make sure it's not obviously broken. - crashtest blackbox without TEST_TMPDIR ``` $ python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest blackbox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest whitebox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py whitebox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none --random_kill_odd=88887 ``` - db_stress without expected_values_dir ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true ``` - db_stress with expected_values_dir and manual corruption ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true --expected_values_dir=./ // modify one byte in "./LATEST.state" $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=false --expected_values_dir=./ ... Verification failed for column family 0 key 0000000000000000 (0): Value not found: NotFound: ... ``` Reviewed By: riversand963 Differential Revision: D30921951 Pulled By: ajkr fbshipit-source-id: babfe218062e55d018c9b046536c0289fb78f41c
3 years ago
expected_values_dir = None
def setup_expected_values_dir():
global expected_values_dir
if expected_values_dir is not None:
return expected_values_dir
expected_dir_prefix = "rocksdb_crashtest_expected_"
test_tmpdir = os.environ.get(_TEST_DIR_ENV_VAR)
if test_tmpdir is None or test_tmpdir == "":
Refactor expected state in stress/crash test (#8913) Summary: This is a precursor refactoring to enable an upcoming feature: persistence failure correctness testing. - Changed `--expected_values_path` to `--expected_values_dir` and migrated "db_crashtest.py" to use the new flag. For persistence failure correctness testing there are multiple possible correct states since unsynced data is allowed to be dropped. Making it possible to restore all these possible correct states will eventually involve files containing snapshots of expected values and DB trace files. - The expected values directory is managed by an `ExpectedStateManager` instance. Managing expected state files is separated out of `SharedState` to prevent `SharedState` from becoming too complex when the new files and features (snapshotting, tracing, and restoring) are introduced. - Migrated expected values file access/management out of `SharedState` into a separate class called `ExpectedState`. This is not exposed directly to the test but rather the `ExpectedState` for the latest values file is accessed via a pass-through API on `ExpectedStateManager`. This forces the test to always access the single latest `ExpectedState`. - Changed the initialization of the latest expected values file to use a tempfile followed by rename, and also add cleanup logic for possible stranded tempfiles. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8913 Test Plan: run in several ways; try to make sure it's not obviously broken. - crashtest blackbox without TEST_TMPDIR ``` $ python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest blackbox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest whitebox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py whitebox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none --random_kill_odd=88887 ``` - db_stress without expected_values_dir ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true ``` - db_stress with expected_values_dir and manual corruption ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true --expected_values_dir=./ // modify one byte in "./LATEST.state" $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=false --expected_values_dir=./ ... Verification failed for column family 0 key 0000000000000000 (0): Value not found: NotFound: ... ``` Reviewed By: riversand963 Differential Revision: D30921951 Pulled By: ajkr fbshipit-source-id: babfe218062e55d018c9b046536c0289fb78f41c
3 years ago
expected_values_dir = tempfile.mkdtemp(
prefix=expected_dir_prefix)
else:
Refactor expected state in stress/crash test (#8913) Summary: This is a precursor refactoring to enable an upcoming feature: persistence failure correctness testing. - Changed `--expected_values_path` to `--expected_values_dir` and migrated "db_crashtest.py" to use the new flag. For persistence failure correctness testing there are multiple possible correct states since unsynced data is allowed to be dropped. Making it possible to restore all these possible correct states will eventually involve files containing snapshots of expected values and DB trace files. - The expected values directory is managed by an `ExpectedStateManager` instance. Managing expected state files is separated out of `SharedState` to prevent `SharedState` from becoming too complex when the new files and features (snapshotting, tracing, and restoring) are introduced. - Migrated expected values file access/management out of `SharedState` into a separate class called `ExpectedState`. This is not exposed directly to the test but rather the `ExpectedState` for the latest values file is accessed via a pass-through API on `ExpectedStateManager`. This forces the test to always access the single latest `ExpectedState`. - Changed the initialization of the latest expected values file to use a tempfile followed by rename, and also add cleanup logic for possible stranded tempfiles. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8913 Test Plan: run in several ways; try to make sure it's not obviously broken. - crashtest blackbox without TEST_TMPDIR ``` $ python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest blackbox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest whitebox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py whitebox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none --random_kill_odd=88887 ``` - db_stress without expected_values_dir ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true ``` - db_stress with expected_values_dir and manual corruption ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true --expected_values_dir=./ // modify one byte in "./LATEST.state" $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=false --expected_values_dir=./ ... Verification failed for column family 0 key 0000000000000000 (0): Value not found: NotFound: ... ``` Reviewed By: riversand963 Differential Revision: D30921951 Pulled By: ajkr fbshipit-source-id: babfe218062e55d018c9b046536c0289fb78f41c
3 years ago
# if tmpdir is specified, store the expected_values_dir under that dir
expected_values_dir = test_tmpdir + "/rocksdb_crashtest_expected"
if os.path.exists(expected_values_dir):
shutil.rmtree(expected_values_dir)
os.mkdir(expected_values_dir)
return expected_values_dir
def is_direct_io_supported(dbname):
with tempfile.NamedTemporaryFile(dir=dbname) as f:
try:
os.open(f.name, os.O_DIRECT)
except BaseException:
return False
return True
blackbox_default_params = {
# total time for this script to test db_stress
"duration": 6000,
# time for one db_stress instance to run
"interval": 120,
# since we will be killing anyway, use large value for ops_per_thread
"ops_per_thread": 100000000,
"set_options_one_in": 10000,
"test_batches_snapshots": 1,
}
whitebox_default_params = {
"duration": 10000,
"log2_keys_per_lock": 10,
"ops_per_thread": 200000,
"random_kill_odd": 888887,
"test_batches_snapshots": lambda: random.randint(0, 1),
}
simple_default_params = {
"allow_concurrent_memtable_write": lambda: random.randint(0, 1),
"column_families": 1,
# TODO: re-enable once below loop succeeds for a while (a few minutes should
# suffice):
# `while rm -rf /dev/shm/single_stress && ./db_stress --clear_column_family_one_in=0 --column_families=1 --db=/dev/shm/single_stress --experimental_mempurge_threshold=5.493146827397074 --flush_one_in=10000 --reopen=0 --write_buffer_size=262144 --value_size_mult=33 --max_write_buffer_number=3 -ops_per_thread=10000; do : ; done`
"experimental_mempurge_threshold": 0,
"max_background_compactions": 1,
"max_bytes_for_level_base": 67108864,
"memtablerep": "skip_list",
"prefixpercent": 0,
"readpercent": 50,
"prefix_size" : -1,
"target_file_size_base": 16777216,
"target_file_size_multiplier": 1,
"test_batches_snapshots": 0,
"write_buffer_size": 32 * 1024 * 1024,
"level_compaction_dynamic_level_bytes": False,
"paranoid_file_checks": lambda: random.choice([0, 1, 1, 1]),
}
blackbox_simple_default_params = {
"open_files": -1,
"set_options_one_in": 0,
}
whitebox_simple_default_params = {}
cf_consistency_params = {
"disable_wal": lambda: random.randint(0, 1),
"reopen": 0,
"test_cf_consistency": 1,
# use small value for write_buffer_size so that RocksDB triggers flush
# more frequently
"write_buffer_size": 1024 * 1024,
"enable_pipelined_write": lambda: random.randint(0, 1),
# Snapshots are used heavily in this test mode, while they are incompatible
# with compaction filter.
"enable_compaction_filter": 0,
}
txn_params = {
"use_txn" : 1,
# Avoid lambda to set it once for the entire test
"txn_write_policy": random.randint(0, 2),
"unordered_write": random.randint(0, 1),
"disable_wal": 0,
# OpenReadOnly after checkpoint is not currnetly compatible with WritePrepared txns
"checkpoint_one_in": 0,
# pipeline write is not currnetly compatible with WritePrepared txns
"enable_pipelined_write": 0,
}
best_efforts_recovery_params = {
"best_efforts_recovery": True,
"skip_verifydb": True,
"verify_db_one_in": 0,
"continuous_verification_interval": 0,
}
blob_params = {
"allow_setting_blob_options_dynamically": 1,
# Enable blob files and GC with a 75% chance initially; note that they might still be
# enabled/disabled during the test via SetOptions
"enable_blob_files": lambda: random.choice([0] + [1] * 3),
"min_blob_size": lambda: random.choice([0, 8, 16]),
"blob_file_size": lambda: random.choice([1048576, 16777216, 268435456, 1073741824]),
"blob_compression_type": lambda: random.choice(["none", "snappy", "lz4", "zstd"]),
"enable_blob_garbage_collection": lambda: random.choice([0] + [1] * 3),
"blob_garbage_collection_age_cutoff": lambda: random.choice([0.0, 0.25, 0.5, 0.75, 1.0]),
Make it possible to force the garbage collection of the oldest blob files (#8994) Summary: The current BlobDB garbage collection logic works by relocating the valid blobs from the oldest blob files as they are encountered during compaction, and cleaning up blob files once they contain nothing but garbage. However, with sufficiently skewed workloads, it is theoretically possible to end up in a situation when few or no compactions get scheduled for the SST files that contain references to the oldest blob files, which can lead to increased space amp due to the lack of GC. In order to efficiently handle such workloads, the patch adds a new BlobDB configuration option called `blob_garbage_collection_force_threshold`, which signals to BlobDB to schedule targeted compactions for the SST files that keep alive the oldest batch of blob files if the overall ratio of garbage in the given blob files meets the threshold *and* all the given blob files are eligible for GC based on `blob_garbage_collection_age_cutoff`. (For example, if the new option is set to 0.9, targeted compactions will get scheduled if the sum of garbage bytes meets or exceeds 90% of the sum of total bytes in the oldest blob files, assuming all affected blob files are below the age-based cutoff.) The net result of these targeted compactions is that the valid blobs in the oldest blob files are relocated and the oldest blob files themselves cleaned up (since *all* SST files that rely on them get compacted away). These targeted compactions are similar to periodic compactions in the sense that they force certain SST files that otherwise would not get picked up to undergo compaction and also in the sense that instead of merging files from multiple levels, they target a single file. (Note: such compactions might still include neighboring files from the same level due to the need of having a "clean cut" boundary but they never include any files from any other level.) This functionality is currently only supported with the leveled compaction style and is inactive by default (since the default value is set to 1.0, i.e. 100%). Pull Request resolved: https://github.com/facebook/rocksdb/pull/8994 Test Plan: Ran `make check` and tested using `db_bench` and the stress/crash tests. Reviewed By: riversand963 Differential Revision: D31489850 Pulled By: ltamasi fbshipit-source-id: 44057d511726a0e2a03c5d9313d7511b3f0c4eab
3 years ago
"blob_garbage_collection_force_threshold": lambda: random.choice([0.5, 0.75, 1.0]),
}
ts_params = {
"test_cf_consistency": 0,
"test_batches_snapshots": 0,
"user_timestamp_size": 8,
"use_merge": 0,
"use_full_merge_v1": 0,
"use_txn": 0,
"read_only": 0,
"secondary_catch_up_one_in": 0,
"continuous_verification_interval": 0,
"checkpoint_one_in": 0,
"enable_blob_files": 0,
"use_blob_db": 0,
"enable_compaction_filter": 0,
"ingest_external_file_one_in": 0,
"use_block_based_filter": 0,
}
def finalize_and_sanitize(src_params):
dest_params = dict([(k, v() if callable(v) else v)
for (k, v) in src_params.items()])
Limit buffering for collecting samples for compression dictionary (#7970) Summary: For dictionary compression, we need to collect some representative samples of the data to be compressed, which we use to either generate or train (when `CompressionOptions::zstd_max_train_bytes > 0`) a dictionary. Previously, the strategy was to buffer all the data blocks during flush, and up to the target file size during compaction. That strategy allowed us to randomly pick samples from as wide a range as possible that'd be guaranteed to land in a single output file. However, some users try to make huge files in memory-constrained environments, where this strategy can cause OOM. This PR introduces an option, `CompressionOptions::max_dict_buffer_bytes`, that limits how much data blocks are buffered before we switch to unbuffered mode (which means creating the per-SST dictionary, writing out the buffered data, and compressing/writing new blocks as soon as they are built). It is not strict as we currently buffer more than just data blocks -- also keys are buffered. But it does make a step towards giving users predictable memory usage. Related changes include: - Changed sampling for dictionary compression to select unique data blocks when there is limited availability of data blocks - Made use of `BlockBuilder::SwapAndReset()` to save an allocation+memcpy when buffering data blocks for building a dictionary - Changed `ParseBoolean()` to accept an input containing characters after the boolean. This is necessary since, with this PR, a value for `CompressionOptions::enabled` is no longer necessarily the final component in the `CompressionOptions` string. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7970 Test Plan: - updated `CompressionOptions` unit tests to verify limit is respected (to the extent expected in the current implementation) in various scenarios of flush/compaction to bottommost/non-bottommost level - looked at jemalloc heap profiles right before and after switching to unbuffered mode during flush/compaction. Verified memory usage in buffering is proportional to the limit set. Reviewed By: pdillinger Differential Revision: D26467994 Pulled By: ajkr fbshipit-source-id: 3da4ef9fba59974e4ef40e40c01611002c861465
4 years ago
if dest_params.get("compression_max_dict_bytes") == 0:
dest_params["compression_zstd_max_train_bytes"] = 0
dest_params["compression_max_dict_buffer_bytes"] = 0
if dest_params.get("compression_type") != "zstd":
dest_params["compression_zstd_max_train_bytes"] = 0
if dest_params.get("allow_concurrent_memtable_write", 1) == 1:
dest_params["memtablerep"] = "skip_list"
if dest_params["mmap_read"] == 1:
dest_params["use_direct_io_for_flush_and_compaction"] = 0
dest_params["use_direct_reads"] = 0
if (dest_params["use_direct_io_for_flush_and_compaction"] == 1
or dest_params["use_direct_reads"] == 1) and \
not is_direct_io_supported(dest_params["db"]):
if is_release_mode():
print("{} does not support direct IO. Disabling use_direct_reads and "
"use_direct_io_for_flush_and_compaction.\n".format(
dest_params["db"]))
dest_params["use_direct_reads"] = 0
dest_params["use_direct_io_for_flush_and_compaction"] = 0
else:
dest_params["mock_direct_io"] = True
# DeleteRange is not currnetly compatible with Txns and timestamp
if (dest_params.get("test_batches_snapshots") == 1 or
dest_params.get("use_txn") == 1 or
dest_params.get("user_timestamp_size") > 0):
dest_params["delpercent"] += dest_params["delrangepercent"]
dest_params["delrangepercent"] = 0
# Only under WritePrepared txns, unordered_write would provide the same guarnatees as vanilla rocksdb
if dest_params.get("unordered_write", 0) == 1:
dest_params["txn_write_policy"] = 1
dest_params["allow_concurrent_memtable_write"] = 1
if dest_params.get("disable_wal", 0) == 1:
dest_params["atomic_flush"] = 1
dest_params["sync"] = 0
dest_params["write_fault_one_in"] = 0
if dest_params.get("open_files", 1) != -1:
# Compaction TTL and periodic compactions are only compatible
# with open_files = -1
dest_params["compaction_ttl"] = 0
dest_params["periodic_compaction_seconds"] = 0
if dest_params.get("compaction_style", 0) == 2:
# Disable compaction TTL in FIFO compaction, because right
# now assertion failures are triggered.
dest_params["compaction_ttl"] = 0
dest_params["periodic_compaction_seconds"] = 0
if dest_params["partition_filters"] == 1:
if dest_params["index_type"] != 2:
dest_params["partition_filters"] = 0
else:
dest_params["use_block_based_filter"] = 0
Add Bloom/Ribbon hybrid API support (#8679) Summary: This is essentially resurrection and fixing of the part of https://github.com/facebook/rocksdb/issues/8198 that was reverted in https://github.com/facebook/rocksdb/issues/8212, using data added in https://github.com/facebook/rocksdb/issues/8246. Basically, when configuring Ribbon filter, you can specify an LSM level before which Bloom will be used instead of Ribbon. But Bloom is only considered for Leveled and Universal compaction styles and file going into a known LSM level. This way, SST file writer, FIFO compaction, etc. use Ribbon filter as you would expect with NewRibbonFilterPolicy. So that this can be controlled with a single int value and so that flushes can be distinguished from intra-L0, we consider flush to go to level -1 for the purposes of this option. (Explained in API comment.) I also expect the most common and recommended Ribbon configuration to use Bloom during flush, to minimize slowing down writes and because according to my estimates, Ribbon only pays off if the structure lives in memory for more than an hour. Thus, I have changed the default for NewRibbonFilterPolicy to be this mild hybrid configuration. I don't really want to add something like NewHybridFilterPolicy because at least the mild hybrid configuration (Bloom for flush, Ribbon otherwise) should be considered a natural choice. C APIs also updated, but because they don't support overloading, rocksdb_filterpolicy_create_ribbon is kept pure ribbon for clarity and rocksdb_filterpolicy_create_ribbon_hybrid must be called for a hybrid configuration. While touching C API, I changed bits per key options from int to double. BuiltinFilterPolicy is needed so that LevelThresholdFilterPolicy doesn't inherit unused fields from BloomFilterPolicy. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8679 Test Plan: new + updated tests, including crash test Reviewed By: jay-zhuang Differential Revision: D30445797 Pulled By: pdillinger fbshipit-source-id: 6f5aeddfd6d79f7e55493b563c2d1d2d568892e1
3 years ago
if dest_params["ribbon_starting_level"] < 999:
dest_params["use_block_based_filter"] = 0
if dest_params.get("atomic_flush", 0) == 1:
# disable pipelined write when atomic flush is used.
dest_params["enable_pipelined_write"] = 0
if dest_params.get("sst_file_manager_bytes_per_sec", 0) == 0:
dest_params["sst_file_manager_bytes_per_truncate"] = 0
if dest_params.get("enable_compaction_filter", 0) == 1:
# Compaction filter is incompatible with snapshots. Need to avoid taking
# snapshots, as well as avoid operations that use snapshots for
# verification.
dest_params["acquire_snapshot_one_in"] = 0
dest_params["compact_range_one_in"] = 0
# Give the iterator ops away to reads.
dest_params["readpercent"] += dest_params.get("iterpercent", 10)
dest_params["iterpercent"] = 0
dest_params["test_batches_snapshots"] = 0
Integrity protection for live updates to WriteBatch (#7748) Summary: This PR adds the foundation classes for key-value integrity protection and the first use case: protecting live updates from the source buffers added to `WriteBatch` through the destination buffer in `MemTable`. The width of the protection info is not yet configurable -- only eight bytes per key is supported. This PR allows users to enable protection by constructing `WriteBatch` with `protection_bytes_per_key == 8`. It does not yet expose a way for users to get integrity protection via other write APIs (e.g., `Put()`, `Merge()`, `Delete()`, etc.). The foundation classes (`ProtectionInfo.*`) embed the coverage info in their type, and provide `Protect.*()` and `Strip.*()` functions to navigate between types with different coverage. For making bytes per key configurable (for powers of two up to eight) in the future, these classes are templated on the unsigned integer type used to store the protection info. That integer contains the XOR'd result of hashes with independent seeds for all covered fields. For integer fields, the hash is computed on the raw unadjusted bytes, so the result is endian-dependent. The most significant bytes are truncated when the hash value (8 bytes) is wider than the protection integer. When `WriteBatch` is constructed with `protection_bytes_per_key == 8`, we hold a `ProtectionInfoKVOTC` (i.e., one that covers key, value, optype aka `ValueType`, timestamp, and CF ID) for each entry added to the batch. The protection info is generated from the original buffers passed by the user, as well as the original metadata generated internally. When writing to memtable, each entry is transformed to a `ProtectionInfoKVOTS` (i.e., dropping coverage of CF ID and adding coverage of sequence number), since at that point we know the sequence number, and have already selected a memtable corresponding to a particular CF. This protection info is verified once the entry is encoded in the `MemTable` buffer. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7748 Test Plan: - an integration test to verify a wide variety of single-byte changes to the encoded `MemTable` buffer are caught - add to stress/crash test to verify it works in variety of configs/operations without intentional corruption - [deferred] unit tests for `ProtectionInfo.*` classes for edge cases like KV swap, `SliceParts` and `Slice` APIs are interchangeable, etc. Reviewed By: pdillinger Differential Revision: D25754492 Pulled By: ajkr fbshipit-source-id: e481bac6c03c2ab268be41359730f1ceb9964866
4 years ago
if dest_params.get("test_batches_snapshots") == 0:
dest_params["batch_protection_bytes_per_key"] = 0
return dest_params
def gen_cmd_params(args):
params = {}
params.update(default_params)
if args.test_type == 'blackbox':
params.update(blackbox_default_params)
if args.test_type == 'whitebox':
params.update(whitebox_default_params)
if args.simple:
params.update(simple_default_params)
if args.test_type == 'blackbox':
params.update(blackbox_simple_default_params)
if args.test_type == 'whitebox':
params.update(whitebox_simple_default_params)
if args.cf_consistency:
params.update(cf_consistency_params)
if args.txn:
params.update(txn_params)
if args.test_best_efforts_recovery:
params.update(best_efforts_recovery_params)
if args.enable_ts:
params.update(ts_params)
# Best-effort recovery and BlobDB are currently incompatible. Test BE recovery
# if specified on the command line; otherwise, apply BlobDB related overrides
# with a 10% chance.
if (not args.test_best_efforts_recovery and
not args.enable_ts and
random.choice([0] * 9 + [1]) == 1):
params.update(blob_params)
for k, v in vars(args).items():
if v is not None:
params[k] = v
return params
def gen_cmd(params, unknown_params):
finalzied_params = finalize_and_sanitize(params)
cmd = [stress_cmd] + [
'--{0}={1}'.format(k, v)
for k, v in [(k, finalzied_params[k]) for k in sorted(finalzied_params)]
if k not in set(['test_type', 'simple', 'duration', 'interval',
'random_kill_odd', 'cf_consistency', 'txn',
'test_best_efforts_recovery', 'enable_ts', 'stress_cmd'])
and v is not None] + unknown_params
return cmd
# Inject inconsistency to db directory.
def inject_inconsistencies_to_db_dir(dir_path):
files = os.listdir(dir_path)
file_num_rgx = re.compile(r'(?P<number>[0-9]{6})')
largest_fnum = 0
for f in files:
m = file_num_rgx.search(f)
if m and not f.startswith('LOG'):
largest_fnum = max(largest_fnum, int(m.group('number')))
candidates = [
f for f in files if re.search(r'[0-9]+\.sst', f)
]
deleted = 0
corrupted = 0
for f in candidates:
rnd = random.randint(0, 99)
f_path = os.path.join(dir_path, f)
if rnd < 10:
os.unlink(f_path)
deleted = deleted + 1
elif 10 <= rnd and rnd < 30:
with open(f_path, "a") as fd:
fd.write('12345678')
corrupted = corrupted + 1
print('Removed %d table files' % deleted)
print('Corrupted %d table files' % corrupted)
# Add corrupted MANIFEST and SST
for num in range(largest_fnum + 1, largest_fnum + 10):
rnd = random.randint(0, 1)
fname = ("MANIFEST-%06d" % num) if rnd == 0 else ("%06d.sst" % num)
print('Write %s' % fname)
with open(os.path.join(dir_path, fname), "w") as fd:
fd.write("garbage")
def execute_cmd(cmd, timeout):
child = subprocess.Popen(cmd, stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
print("Running db_stress with pid=%d: %s\n\n"
% (child.pid, ' '.join(cmd)))
try:
outs, errs = child.communicate(timeout=timeout)
hit_timeout = False
print("WARNING: db_stress ended before kill: exitcode=%d\n"
% child.returncode)
except subprocess.TimeoutExpired:
hit_timeout = True
child.kill()
print("KILLED %d\n" % child.pid)
outs, errs = child.communicate()
return hit_timeout, child.returncode, outs.decode('utf-8'), errs.decode('utf-8')
# This script runs and kills db_stress multiple times. It checks consistency
# in case of unsafe crashes in RocksDB.
def blackbox_crash_main(args, unknown_args):
cmd_params = gen_cmd_params(args)
dbname = get_dbname('blackbox')
exit_time = time.time() + cmd_params['duration']
print("Running blackbox-crash-test with \n"
+ "interval_between_crash=" + str(cmd_params['interval']) + "\n"
+ "total-duration=" + str(cmd_params['duration']) + "\n")
while time.time() < exit_time:
cmd = gen_cmd(dict(
list(cmd_params.items())
+ list({'db': dbname}.items())), unknown_args)
hit_timeout, retcode, outs, errs = execute_cmd(cmd, cmd_params['interval'])
if not hit_timeout:
print('Exit Before Killing')
print('stdout:')
print(outs)
print('stderr:')
print(errs)
sys.exit(2)
for line in errs.split('\n'):
if line != '' and not line.startswith('WARNING'):
print('stderr has error message:')
print('***' + line + '***')
time.sleep(1) # time to stabilize before the next run
if args.test_best_efforts_recovery:
inject_inconsistencies_to_db_dir(dbname)
time.sleep(1) # time to stabilize before the next run
# we need to clean up after ourselves -- only do this on test success
shutil.rmtree(dbname, True)
# This python script runs db_stress multiple times. Some runs with
# kill_random_test that causes rocksdb to crash at various points in code.
def whitebox_crash_main(args, unknown_args):
cmd_params = gen_cmd_params(args)
dbname = get_dbname('whitebox')
cur_time = time.time()
exit_time = cur_time + cmd_params['duration']
half_time = cur_time + cmd_params['duration'] // 2
print("Running whitebox-crash-test with \n"
+ "total-duration=" + str(cmd_params['duration']) + "\n")
total_check_mode = 4
check_mode = 0
kill_random_test = cmd_params['random_kill_odd']
kill_mode = 0
while time.time() < exit_time:
if check_mode == 0:
additional_opts = {
# use large ops per thread since we will kill it anyway
"ops_per_thread": 100 * cmd_params['ops_per_thread'],
}
# run with kill_random_test, with three modes.
# Mode 0 covers all kill points. Mode 1 covers less kill points but
# increases change of triggering them. Mode 2 covers even less
# frequent kill points and further increases triggering change.
if kill_mode == 0:
additional_opts.update({
"kill_random_test": kill_random_test,
})
elif kill_mode == 1:
if cmd_params.get('disable_wal', 0) == 1:
my_kill_odd = kill_random_test // 50 + 1
else:
my_kill_odd = kill_random_test // 10 + 1
additional_opts.update({
"kill_random_test": my_kill_odd,
"kill_exclude_prefixes": "WritableFileWriter::Append,"
+ "WritableFileWriter::WriteBuffered",
})
elif kill_mode == 2:
# TODO: May need to adjust random odds if kill_random_test
# is too small.
additional_opts.update({
"kill_random_test": (kill_random_test // 5000 + 1),
"kill_exclude_prefixes": "WritableFileWriter::Append,"
"WritableFileWriter::WriteBuffered,"
"PosixMmapFile::Allocate,WritableFileWriter::Flush",
})
# Run kill mode 0, 1 and 2 by turn.
kill_mode = (kill_mode + 1) % 3
elif check_mode == 1:
# normal run with universal compaction mode
additional_opts = {
"kill_random_test": None,
"ops_per_thread": cmd_params['ops_per_thread'],
"compaction_style": 1,
}
# Single level universal has a lot of special logic. Ensure we cover
# it sometimes.
if random.randint(0, 1) == 1:
additional_opts.update({
"num_levels": 1,
})
elif check_mode == 2:
# normal run with FIFO compaction mode
# ops_per_thread is divided by 5 because FIFO compaction
# style is quite a bit slower on reads with lot of files
additional_opts = {
"kill_random_test": None,
"ops_per_thread": cmd_params['ops_per_thread'] // 5,
"compaction_style": 2,
}
else:
# normal run
additional_opts = {
"kill_random_test": None,
"ops_per_thread": cmd_params['ops_per_thread'],
}
cmd = gen_cmd(dict(list(cmd_params.items())
+ list(additional_opts.items())
+ list({'db': dbname}.items())), unknown_args)
print("Running:" + ' '.join(cmd) + "\n") # noqa: E999 T25377293 Grandfathered in
# If the running time is 15 minutes over the run time, explicit kill and
# exit even if white box kill didn't hit. This is to guarantee run time
# limit, as if it runs as a job, running too long will create problems
# for job scheduling or execution.
# TODO detect a hanging condition. The job might run too long as RocksDB
# hits a hanging bug.
hit_timeout, retncode, stdoutdata, stderrdata = execute_cmd(
cmd, exit_time - time.time() + 900)
msg = ("check_mode={0}, kill option={1}, exitcode={2}\n".format(
check_mode, additional_opts['kill_random_test'], retncode))
print(msg)
print(stdoutdata)
print(stderrdata)
if hit_timeout:
print("Killing the run for running too long")
break
expected = False
if additional_opts['kill_random_test'] is None and (retncode == 0):
# we expect zero retncode if no kill option
expected = True
elif additional_opts['kill_random_test'] is not None and retncode <= 0:
# When kill option is given, the test MIGHT kill itself.
# If it does, negative retncode is expected. Otherwise 0.
expected = True
if not expected:
print("TEST FAILED. See kill option and exit code above!!!\n")
sys.exit(1)
stderrdata = stderrdata.lower()
errorcount = (stderrdata.count('error') -
stderrdata.count('got errors 0 times'))
print("#times error occurred in output is " + str(errorcount) +
"\n")
if (errorcount > 0):
print("TEST FAILED. Output has 'error'!!!\n")
sys.exit(2)
if (stderrdata.find('fail') >= 0):
print("TEST FAILED. Output has 'fail'!!!\n")
sys.exit(2)
# First half of the duration, keep doing kill test. For the next half,
# try different modes.
if time.time() > half_time:
# we need to clean up after ourselves -- only do this on test
# success
shutil.rmtree(dbname, True)
os.mkdir(dbname)
Refactor expected state in stress/crash test (#8913) Summary: This is a precursor refactoring to enable an upcoming feature: persistence failure correctness testing. - Changed `--expected_values_path` to `--expected_values_dir` and migrated "db_crashtest.py" to use the new flag. For persistence failure correctness testing there are multiple possible correct states since unsynced data is allowed to be dropped. Making it possible to restore all these possible correct states will eventually involve files containing snapshots of expected values and DB trace files. - The expected values directory is managed by an `ExpectedStateManager` instance. Managing expected state files is separated out of `SharedState` to prevent `SharedState` from becoming too complex when the new files and features (snapshotting, tracing, and restoring) are introduced. - Migrated expected values file access/management out of `SharedState` into a separate class called `ExpectedState`. This is not exposed directly to the test but rather the `ExpectedState` for the latest values file is accessed via a pass-through API on `ExpectedStateManager`. This forces the test to always access the single latest `ExpectedState`. - Changed the initialization of the latest expected values file to use a tempfile followed by rename, and also add cleanup logic for possible stranded tempfiles. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8913 Test Plan: run in several ways; try to make sure it's not obviously broken. - crashtest blackbox without TEST_TMPDIR ``` $ python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest blackbox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest whitebox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py whitebox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none --random_kill_odd=88887 ``` - db_stress without expected_values_dir ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true ``` - db_stress with expected_values_dir and manual corruption ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true --expected_values_dir=./ // modify one byte in "./LATEST.state" $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=false --expected_values_dir=./ ... Verification failed for column family 0 key 0000000000000000 (0): Value not found: NotFound: ... ``` Reviewed By: riversand963 Differential Revision: D30921951 Pulled By: ajkr fbshipit-source-id: babfe218062e55d018c9b046536c0289fb78f41c
3 years ago
cmd_params.pop('expected_values_dir', None)
check_mode = (check_mode + 1) % total_check_mode
time.sleep(1) # time to stabilize after a kill
def main():
global stress_cmd
parser = argparse.ArgumentParser(description="This script runs and kills \
db_stress multiple times")
parser.add_argument("test_type", choices=["blackbox", "whitebox"])
parser.add_argument("--simple", action="store_true")
parser.add_argument("--cf_consistency", action='store_true')
parser.add_argument("--txn", action='store_true')
parser.add_argument("--test_best_efforts_recovery", action='store_true')
parser.add_argument("--enable_ts", action='store_true')
parser.add_argument("--stress_cmd")
all_params = dict(list(default_params.items())
+ list(blackbox_default_params.items())
+ list(whitebox_default_params.items())
+ list(simple_default_params.items())
+ list(blackbox_simple_default_params.items())
+ list(whitebox_simple_default_params.items())
+ list(blob_params.items())
+ list(ts_params.items()))
for k, v in all_params.items():
parser.add_argument("--" + k, type=type(v() if callable(v) else v))
# unknown_args are passed directly to db_stress
args, unknown_args = parser.parse_known_args()
test_tmpdir = os.environ.get(_TEST_DIR_ENV_VAR)
if test_tmpdir is not None and not os.path.isdir(test_tmpdir):
print('%s env var is set to a non-existent directory: %s' %
(_TEST_DIR_ENV_VAR, test_tmpdir))
sys.exit(1)
if args.stress_cmd:
stress_cmd = args.stress_cmd
if args.test_type == 'blackbox':
blackbox_crash_main(args, unknown_args)
if args.test_type == 'whitebox':
whitebox_crash_main(args, unknown_args)
Refactor expected state in stress/crash test (#8913) Summary: This is a precursor refactoring to enable an upcoming feature: persistence failure correctness testing. - Changed `--expected_values_path` to `--expected_values_dir` and migrated "db_crashtest.py" to use the new flag. For persistence failure correctness testing there are multiple possible correct states since unsynced data is allowed to be dropped. Making it possible to restore all these possible correct states will eventually involve files containing snapshots of expected values and DB trace files. - The expected values directory is managed by an `ExpectedStateManager` instance. Managing expected state files is separated out of `SharedState` to prevent `SharedState` from becoming too complex when the new files and features (snapshotting, tracing, and restoring) are introduced. - Migrated expected values file access/management out of `SharedState` into a separate class called `ExpectedState`. This is not exposed directly to the test but rather the `ExpectedState` for the latest values file is accessed via a pass-through API on `ExpectedStateManager`. This forces the test to always access the single latest `ExpectedState`. - Changed the initialization of the latest expected values file to use a tempfile followed by rename, and also add cleanup logic for possible stranded tempfiles. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8913 Test Plan: run in several ways; try to make sure it's not obviously broken. - crashtest blackbox without TEST_TMPDIR ``` $ python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest blackbox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py blackbox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none ``` - crashtest whitebox with TEST_TMPDIR ``` $ TEST_TMPDIR=/dev/shm python3 tools/db_crashtest.py whitebox --simple --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --duration=120 --interval=10 --compression_type=none --blob_compression_type=none --random_kill_odd=88887 ``` - db_stress without expected_values_dir ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true ``` - db_stress with expected_values_dir and manual corruption ``` $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=true --expected_values_dir=./ // modify one byte in "./LATEST.state" $ ./db_stress --write_buffer_size=1048576 --target_file_size_base=1048576 --max_bytes_for_level_base=4194304 --max_key=100000 --value_size_mult=33 --compression_type=none --ops_per_thread=10000 --clear_column_family_one_in=0 --destroy_db_initially=false --expected_values_dir=./ ... Verification failed for column family 0 key 0000000000000000 (0): Value not found: NotFound: ... ``` Reviewed By: riversand963 Differential Revision: D30921951 Pulled By: ajkr fbshipit-source-id: babfe218062e55d018c9b046536c0289fb78f41c
3 years ago
# Only delete the `expected_values_dir` if test passes
if expected_values_dir is not None:
shutil.rmtree(expected_values_dir)
if __name__ == '__main__':
main()