rocksdb/utilities/fault_injection_fs.h

631 lines
23 KiB
C
Raw Normal View History

// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
//
// Copyright 2014 The LevelDB Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file. See the AUTHORS file for names of contributors.
// This test uses a custom FileSystem to keep track of the state of a file
// system the last "Sync". The data being written is cached in a "buffer".
// Only when "Sync" is called, the data will be persistent. It can similate
// file data loss (or entire files) not protected by a "Sync". For any of the
// FileSystem related operations, by specify the "IOStatus Error", a specific
// error can be returned when file system is not activated.
#pragma once
#include <algorithm>
#include <map>
#include <set>
#include <string>
#include "file/filename.h"
#include "rocksdb/file_system.h"
#include "util/mutexlock.h"
#include "util/random.h"
#include "util/thread_local.h"
namespace ROCKSDB_NAMESPACE {
class TestFSWritableFile;
class FaultInjectionTestFS;
struct FSFileState {
std::string filename_;
Enable reading un-synced data in db stress test (#12752) Summary: **Context/Summary:** There are a few blockers to enabling reading un-synced data in db stress test (1) GetFileSize() will always return 0 for file written under direct IO because we don't track the last flushed position for `TestFSWritableFile` under direct IO. So it will surface as ``` Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` (2) A couple minor FIXME in left in https://github.com/facebook/rocksdb/pull/12729. This PR fixed (1) and (2) and enabled reading un-synced data in stress test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12752 Test Plan: - The following command failed before this PR and passed after. ``` ./db_stress --WAL_size_limit_MB=1 --WAL_ttl_seconds=0 --acquire_snapshot_one_in=100 --adaptive_readahead=0 --adm_policy=0 --advise_random_on_open=1 --allow_concurrent_memtable_write=0 --allow_data_in_errors=True --allow_fallocate=1 --async_io=1 --atomic_flush=1 --auto_readahead_size=1 --avoid_flush_during_recovery=0 --avoid_flush_during_shutdown=1 --avoid_unnecessary_blocking_io=0 --backup_max_size=104857600 --backup_one_in=100000 --batch_protection_bytes_per_key=0 --bgerror_resume_retry_interval=10000 --block_align=0 --block_protection_bytes_per_key=4 --block_size=16384 --bloom_before_level=2147483647 --bloom_bits=37.92024930098943 --bottommost_compression_type=disable --bottommost_file_compaction_delay=0 --bytes_per_sync=0 --cache_index_and_filter_blocks=1 --cache_index_and_filter_blocks_with_high_priority=0 --cache_size=8388608 --cache_type=auto_hyper_clock_cache --charge_compression_dictionary_building_buffer=1 --charge_file_metadata=0 --charge_filter_construction=0 --charge_table_reader=1 --check_multiget_consistency=0 --check_multiget_entity_consistency=0 --checkpoint_one_in=1000000 --checksum_type=kXXH3 --clear_column_family_one_in=0 --column_families=1 --compact_files_one_in=1000 --compact_range_one_in=1000 --compaction_pri=3 --compaction_readahead_size=0 --compaction_ttl=10 --compress_format_version=2 --compressed_secondary_cache_size=8388608 --compression_checksum=1 --compression_max_dict_buffer_bytes=0 --compression_max_dict_bytes=0 --compression_parallel_threads=1 --compression_type=zlib --compression_use_zstd_dict_trainer=0 --compression_zstd_max_train_bytes=0 --continuous_verification_interval=0 --daily_offpeak_time_utc= --data_block_index_type=0 --db=/tmp/rocksdb_crashtest_blackbox4deg_c5e --db_write_buffer_size=0 --default_temperature=kWarm --default_write_temperature=kHot --delete_obsolete_files_period_micros=30000000 --delpercent=4 --delrangepercent=1 --destroy_db_initially=0 --detect_filter_construct_corruption=0 --disable_file_deletions_one_in=1000000 --disable_manual_compaction_one_in=10000 --disable_wal=1 --dump_malloc_stats=0 --enable_checksum_handoff=0 --enable_compaction_filter=0 --enable_custom_split_merge=1 --enable_do_not_compress_roles=0 --enable_index_compression=1 --enable_memtable_insert_with_hint_prefix_extractor=0 --enable_pipelined_write=0 --enable_sst_partitioner_factory=1 --enable_thread_tracking=0 --enable_write_thread_adaptive_yield=0 --expected_values_dir=/tmp/rocksdb_crashtest_expected_8whyhdxm --fail_if_options_file_error=0 --fifo_allow_compaction=1 --file_checksum_impl=xxh64 --fill_cache=1 --flush_one_in=1000 --format_version=4 --get_all_column_family_metadata_one_in=10000 --get_current_wal_file_one_in=0 --get_live_files_apis_one_in=1000000 --get_properties_of_all_tables_one_in=100000 --get_property_one_in=100000 --get_sorted_wal_files_one_in=0 --hard_pending_compaction_bytes_limit=274877906944 --high_pri_pool_ratio=0.5 --index_block_restart_interval=9 --index_shortening=0 --index_type=0 --ingest_external_file_one_in=0 --initial_auto_readahead_size=0 --inplace_update_support=0 --iterpercent=10 --key_len_percent_dist=1,30,69 --key_may_exist_one_in=100000 --last_level_temperature=kUnknown --level_compaction_dynamic_level_bytes=1 --lock_wal_one_in=100 --log_file_time_to_roll=0 --log_readahead_size=0 --long_running_snapshots=0 --low_pri_pool_ratio=0 --lowest_used_cache_tier=0 --manifest_preallocation_size=5120 --manual_wal_flush_one_in=0 --mark_for_compaction_one_file_in=0 --max_auto_readahead_size=0 --max_background_compactions=1 --max_bytes_for_level_base=67108864 --max_key=100000 --max_key_len=3 --max_log_file_size=0 --max_manifest_file_size=1073741824 --max_sequential_skip_in_iterations=2 --max_total_wal_size=0 --max_write_batch_group_size_bytes=1048576 --max_write_buffer_number=3 --max_write_buffer_size_to_maintain=2097152 --memtable_insert_hint_per_batch=1 --memtable_max_range_deletions=100 --memtable_prefix_bloom_size_ratio=0.001 --memtable_protection_bytes_per_key=0 --memtable_whole_key_filtering=0 --memtablerep=skip_list --metadata_charge_policy=1 --min_write_buffer_number_to_merge=2 --mmap_read=0 --mock_direct_io=False --nooverwritepercent=1 --num_file_reads_for_auto_readahead=1 --open_files=-1 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=16 --ops_per_thread=100000000 --optimize_filters_for_hits=1 --optimize_filters_for_memory=0 --optimize_multiget_for_io=0 --paranoid_file_checks=1 --partition_filters=0 --partition_pinning=1 --pause_background_one_in=1000000 --periodic_compaction_seconds=1000 --prefix_size=5 --prefixpercent=5 --prepopulate_block_cache=1 --preserve_internal_time_seconds=0 --progress_reports=0 --promote_l0_one_in=0 --read_amp_bytes_per_bit=0 --read_fault_one_in=1000 --readahead_size=16384 --readpercent=45 --recycle_log_file_num=0 --reopen=0 --report_bg_io_stats=0 --reset_stats_one_in=1000000 --sample_for_compression=5 --secondary_cache_fault_one_in=32 --secondary_cache_uri= --set_options_one_in=0 --skip_stats_update_on_db_open=0 --snapshot_hold_ops=100000 --soft_pending_compaction_bytes_limit=68719476736 --sst_file_manager_bytes_per_sec=104857600 --sst_file_manager_bytes_per_truncate=1048576 --stats_dump_period_sec=0 --stats_history_buffer_size=0 --strict_bytes_per_sync=0 --subcompactions=2 --sync=0 --sync_fault_injection=1 --table_cache_numshardbits=0 --target_file_size_base=16777216 --target_file_size_multiplier=1 --test_batches_snapshots=0 --top_level_index_pinning=0 --uncache_aggressiveness=1 --universal_max_read_amp=-1 --unpartitioned_pinning=0 --use_adaptive_mutex=0 --use_adaptive_mutex_lru=0 --use_attribute_group=1 --use_delta_encoding=1 --use_direct_io_for_flush_and_compaction=1 --use_direct_reads=0 --use_full_merge_v1=0 --use_get_entity=0 --use_merge=0 --use_multi_cf_iterator=0 --use_multi_get_entity=1 --use_multiget=0 --use_put_entity_one_in=5 --use_timed_put_one_in=0 --use_write_buffer_manager=0 --user_timestamp_size=0 --value_size_mult=32 --verification_only=0 --verify_checksum=1 --verify_checksum_one_in=10 --verify_compression=1 --verify_db_one_in=10000 --verify_file_checksums_one_in=10 --verify_iterator_with_expected_state_one_in=5 --verify_sst_unique_id_in_manifest=1 --wal_bytes_per_sync=0 --wal_compression=zstd --write_buffer_size=33554432 --write_dbid_to_manifest=0 --write_fault_one_in=0 --writepercent=35 Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` - Run python3 tools/db_crashtest.py --simple blackbox --lock_wal_one_in=10 --backup_one_in=10 --sync_fault_injection=0 --use_direct_io_for_flush_and_compaction=0 for 1 hour - Monitor stress test CI Reviewed By: pdillinger Differential Revision: D58395807 Pulled By: hx235 fbshipit-source-id: 7d4b321acc0a0af3501b62dc417a7f6e2d318265
2024-06-18 21:41:14 +00:00
ssize_t pos_at_last_append_;
ssize_t pos_at_last_sync_;
std::string buffer_;
explicit FSFileState(const std::string& filename)
Enable reading un-synced data in db stress test (#12752) Summary: **Context/Summary:** There are a few blockers to enabling reading un-synced data in db stress test (1) GetFileSize() will always return 0 for file written under direct IO because we don't track the last flushed position for `TestFSWritableFile` under direct IO. So it will surface as ``` Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` (2) A couple minor FIXME in left in https://github.com/facebook/rocksdb/pull/12729. This PR fixed (1) and (2) and enabled reading un-synced data in stress test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12752 Test Plan: - The following command failed before this PR and passed after. ``` ./db_stress --WAL_size_limit_MB=1 --WAL_ttl_seconds=0 --acquire_snapshot_one_in=100 --adaptive_readahead=0 --adm_policy=0 --advise_random_on_open=1 --allow_concurrent_memtable_write=0 --allow_data_in_errors=True --allow_fallocate=1 --async_io=1 --atomic_flush=1 --auto_readahead_size=1 --avoid_flush_during_recovery=0 --avoid_flush_during_shutdown=1 --avoid_unnecessary_blocking_io=0 --backup_max_size=104857600 --backup_one_in=100000 --batch_protection_bytes_per_key=0 --bgerror_resume_retry_interval=10000 --block_align=0 --block_protection_bytes_per_key=4 --block_size=16384 --bloom_before_level=2147483647 --bloom_bits=37.92024930098943 --bottommost_compression_type=disable --bottommost_file_compaction_delay=0 --bytes_per_sync=0 --cache_index_and_filter_blocks=1 --cache_index_and_filter_blocks_with_high_priority=0 --cache_size=8388608 --cache_type=auto_hyper_clock_cache --charge_compression_dictionary_building_buffer=1 --charge_file_metadata=0 --charge_filter_construction=0 --charge_table_reader=1 --check_multiget_consistency=0 --check_multiget_entity_consistency=0 --checkpoint_one_in=1000000 --checksum_type=kXXH3 --clear_column_family_one_in=0 --column_families=1 --compact_files_one_in=1000 --compact_range_one_in=1000 --compaction_pri=3 --compaction_readahead_size=0 --compaction_ttl=10 --compress_format_version=2 --compressed_secondary_cache_size=8388608 --compression_checksum=1 --compression_max_dict_buffer_bytes=0 --compression_max_dict_bytes=0 --compression_parallel_threads=1 --compression_type=zlib --compression_use_zstd_dict_trainer=0 --compression_zstd_max_train_bytes=0 --continuous_verification_interval=0 --daily_offpeak_time_utc= --data_block_index_type=0 --db=/tmp/rocksdb_crashtest_blackbox4deg_c5e --db_write_buffer_size=0 --default_temperature=kWarm --default_write_temperature=kHot --delete_obsolete_files_period_micros=30000000 --delpercent=4 --delrangepercent=1 --destroy_db_initially=0 --detect_filter_construct_corruption=0 --disable_file_deletions_one_in=1000000 --disable_manual_compaction_one_in=10000 --disable_wal=1 --dump_malloc_stats=0 --enable_checksum_handoff=0 --enable_compaction_filter=0 --enable_custom_split_merge=1 --enable_do_not_compress_roles=0 --enable_index_compression=1 --enable_memtable_insert_with_hint_prefix_extractor=0 --enable_pipelined_write=0 --enable_sst_partitioner_factory=1 --enable_thread_tracking=0 --enable_write_thread_adaptive_yield=0 --expected_values_dir=/tmp/rocksdb_crashtest_expected_8whyhdxm --fail_if_options_file_error=0 --fifo_allow_compaction=1 --file_checksum_impl=xxh64 --fill_cache=1 --flush_one_in=1000 --format_version=4 --get_all_column_family_metadata_one_in=10000 --get_current_wal_file_one_in=0 --get_live_files_apis_one_in=1000000 --get_properties_of_all_tables_one_in=100000 --get_property_one_in=100000 --get_sorted_wal_files_one_in=0 --hard_pending_compaction_bytes_limit=274877906944 --high_pri_pool_ratio=0.5 --index_block_restart_interval=9 --index_shortening=0 --index_type=0 --ingest_external_file_one_in=0 --initial_auto_readahead_size=0 --inplace_update_support=0 --iterpercent=10 --key_len_percent_dist=1,30,69 --key_may_exist_one_in=100000 --last_level_temperature=kUnknown --level_compaction_dynamic_level_bytes=1 --lock_wal_one_in=100 --log_file_time_to_roll=0 --log_readahead_size=0 --long_running_snapshots=0 --low_pri_pool_ratio=0 --lowest_used_cache_tier=0 --manifest_preallocation_size=5120 --manual_wal_flush_one_in=0 --mark_for_compaction_one_file_in=0 --max_auto_readahead_size=0 --max_background_compactions=1 --max_bytes_for_level_base=67108864 --max_key=100000 --max_key_len=3 --max_log_file_size=0 --max_manifest_file_size=1073741824 --max_sequential_skip_in_iterations=2 --max_total_wal_size=0 --max_write_batch_group_size_bytes=1048576 --max_write_buffer_number=3 --max_write_buffer_size_to_maintain=2097152 --memtable_insert_hint_per_batch=1 --memtable_max_range_deletions=100 --memtable_prefix_bloom_size_ratio=0.001 --memtable_protection_bytes_per_key=0 --memtable_whole_key_filtering=0 --memtablerep=skip_list --metadata_charge_policy=1 --min_write_buffer_number_to_merge=2 --mmap_read=0 --mock_direct_io=False --nooverwritepercent=1 --num_file_reads_for_auto_readahead=1 --open_files=-1 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=16 --ops_per_thread=100000000 --optimize_filters_for_hits=1 --optimize_filters_for_memory=0 --optimize_multiget_for_io=0 --paranoid_file_checks=1 --partition_filters=0 --partition_pinning=1 --pause_background_one_in=1000000 --periodic_compaction_seconds=1000 --prefix_size=5 --prefixpercent=5 --prepopulate_block_cache=1 --preserve_internal_time_seconds=0 --progress_reports=0 --promote_l0_one_in=0 --read_amp_bytes_per_bit=0 --read_fault_one_in=1000 --readahead_size=16384 --readpercent=45 --recycle_log_file_num=0 --reopen=0 --report_bg_io_stats=0 --reset_stats_one_in=1000000 --sample_for_compression=5 --secondary_cache_fault_one_in=32 --secondary_cache_uri= --set_options_one_in=0 --skip_stats_update_on_db_open=0 --snapshot_hold_ops=100000 --soft_pending_compaction_bytes_limit=68719476736 --sst_file_manager_bytes_per_sec=104857600 --sst_file_manager_bytes_per_truncate=1048576 --stats_dump_period_sec=0 --stats_history_buffer_size=0 --strict_bytes_per_sync=0 --subcompactions=2 --sync=0 --sync_fault_injection=1 --table_cache_numshardbits=0 --target_file_size_base=16777216 --target_file_size_multiplier=1 --test_batches_snapshots=0 --top_level_index_pinning=0 --uncache_aggressiveness=1 --universal_max_read_amp=-1 --unpartitioned_pinning=0 --use_adaptive_mutex=0 --use_adaptive_mutex_lru=0 --use_attribute_group=1 --use_delta_encoding=1 --use_direct_io_for_flush_and_compaction=1 --use_direct_reads=0 --use_full_merge_v1=0 --use_get_entity=0 --use_merge=0 --use_multi_cf_iterator=0 --use_multi_get_entity=1 --use_multiget=0 --use_put_entity_one_in=5 --use_timed_put_one_in=0 --use_write_buffer_manager=0 --user_timestamp_size=0 --value_size_mult=32 --verification_only=0 --verify_checksum=1 --verify_checksum_one_in=10 --verify_compression=1 --verify_db_one_in=10000 --verify_file_checksums_one_in=10 --verify_iterator_with_expected_state_one_in=5 --verify_sst_unique_id_in_manifest=1 --wal_bytes_per_sync=0 --wal_compression=zstd --write_buffer_size=33554432 --write_dbid_to_manifest=0 --write_fault_one_in=0 --writepercent=35 Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` - Run python3 tools/db_crashtest.py --simple blackbox --lock_wal_one_in=10 --backup_one_in=10 --sync_fault_injection=0 --use_direct_io_for_flush_and_compaction=0 for 1 hour - Monitor stress test CI Reviewed By: pdillinger Differential Revision: D58395807 Pulled By: hx235 fbshipit-source-id: 7d4b321acc0a0af3501b62dc417a7f6e2d318265
2024-06-18 21:41:14 +00:00
: filename_(filename), pos_at_last_append_(-1), pos_at_last_sync_(-1) {}
Enable reading un-synced data in db stress test (#12752) Summary: **Context/Summary:** There are a few blockers to enabling reading un-synced data in db stress test (1) GetFileSize() will always return 0 for file written under direct IO because we don't track the last flushed position for `TestFSWritableFile` under direct IO. So it will surface as ``` Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` (2) A couple minor FIXME in left in https://github.com/facebook/rocksdb/pull/12729. This PR fixed (1) and (2) and enabled reading un-synced data in stress test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12752 Test Plan: - The following command failed before this PR and passed after. ``` ./db_stress --WAL_size_limit_MB=1 --WAL_ttl_seconds=0 --acquire_snapshot_one_in=100 --adaptive_readahead=0 --adm_policy=0 --advise_random_on_open=1 --allow_concurrent_memtable_write=0 --allow_data_in_errors=True --allow_fallocate=1 --async_io=1 --atomic_flush=1 --auto_readahead_size=1 --avoid_flush_during_recovery=0 --avoid_flush_during_shutdown=1 --avoid_unnecessary_blocking_io=0 --backup_max_size=104857600 --backup_one_in=100000 --batch_protection_bytes_per_key=0 --bgerror_resume_retry_interval=10000 --block_align=0 --block_protection_bytes_per_key=4 --block_size=16384 --bloom_before_level=2147483647 --bloom_bits=37.92024930098943 --bottommost_compression_type=disable --bottommost_file_compaction_delay=0 --bytes_per_sync=0 --cache_index_and_filter_blocks=1 --cache_index_and_filter_blocks_with_high_priority=0 --cache_size=8388608 --cache_type=auto_hyper_clock_cache --charge_compression_dictionary_building_buffer=1 --charge_file_metadata=0 --charge_filter_construction=0 --charge_table_reader=1 --check_multiget_consistency=0 --check_multiget_entity_consistency=0 --checkpoint_one_in=1000000 --checksum_type=kXXH3 --clear_column_family_one_in=0 --column_families=1 --compact_files_one_in=1000 --compact_range_one_in=1000 --compaction_pri=3 --compaction_readahead_size=0 --compaction_ttl=10 --compress_format_version=2 --compressed_secondary_cache_size=8388608 --compression_checksum=1 --compression_max_dict_buffer_bytes=0 --compression_max_dict_bytes=0 --compression_parallel_threads=1 --compression_type=zlib --compression_use_zstd_dict_trainer=0 --compression_zstd_max_train_bytes=0 --continuous_verification_interval=0 --daily_offpeak_time_utc= --data_block_index_type=0 --db=/tmp/rocksdb_crashtest_blackbox4deg_c5e --db_write_buffer_size=0 --default_temperature=kWarm --default_write_temperature=kHot --delete_obsolete_files_period_micros=30000000 --delpercent=4 --delrangepercent=1 --destroy_db_initially=0 --detect_filter_construct_corruption=0 --disable_file_deletions_one_in=1000000 --disable_manual_compaction_one_in=10000 --disable_wal=1 --dump_malloc_stats=0 --enable_checksum_handoff=0 --enable_compaction_filter=0 --enable_custom_split_merge=1 --enable_do_not_compress_roles=0 --enable_index_compression=1 --enable_memtable_insert_with_hint_prefix_extractor=0 --enable_pipelined_write=0 --enable_sst_partitioner_factory=1 --enable_thread_tracking=0 --enable_write_thread_adaptive_yield=0 --expected_values_dir=/tmp/rocksdb_crashtest_expected_8whyhdxm --fail_if_options_file_error=0 --fifo_allow_compaction=1 --file_checksum_impl=xxh64 --fill_cache=1 --flush_one_in=1000 --format_version=4 --get_all_column_family_metadata_one_in=10000 --get_current_wal_file_one_in=0 --get_live_files_apis_one_in=1000000 --get_properties_of_all_tables_one_in=100000 --get_property_one_in=100000 --get_sorted_wal_files_one_in=0 --hard_pending_compaction_bytes_limit=274877906944 --high_pri_pool_ratio=0.5 --index_block_restart_interval=9 --index_shortening=0 --index_type=0 --ingest_external_file_one_in=0 --initial_auto_readahead_size=0 --inplace_update_support=0 --iterpercent=10 --key_len_percent_dist=1,30,69 --key_may_exist_one_in=100000 --last_level_temperature=kUnknown --level_compaction_dynamic_level_bytes=1 --lock_wal_one_in=100 --log_file_time_to_roll=0 --log_readahead_size=0 --long_running_snapshots=0 --low_pri_pool_ratio=0 --lowest_used_cache_tier=0 --manifest_preallocation_size=5120 --manual_wal_flush_one_in=0 --mark_for_compaction_one_file_in=0 --max_auto_readahead_size=0 --max_background_compactions=1 --max_bytes_for_level_base=67108864 --max_key=100000 --max_key_len=3 --max_log_file_size=0 --max_manifest_file_size=1073741824 --max_sequential_skip_in_iterations=2 --max_total_wal_size=0 --max_write_batch_group_size_bytes=1048576 --max_write_buffer_number=3 --max_write_buffer_size_to_maintain=2097152 --memtable_insert_hint_per_batch=1 --memtable_max_range_deletions=100 --memtable_prefix_bloom_size_ratio=0.001 --memtable_protection_bytes_per_key=0 --memtable_whole_key_filtering=0 --memtablerep=skip_list --metadata_charge_policy=1 --min_write_buffer_number_to_merge=2 --mmap_read=0 --mock_direct_io=False --nooverwritepercent=1 --num_file_reads_for_auto_readahead=1 --open_files=-1 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=16 --ops_per_thread=100000000 --optimize_filters_for_hits=1 --optimize_filters_for_memory=0 --optimize_multiget_for_io=0 --paranoid_file_checks=1 --partition_filters=0 --partition_pinning=1 --pause_background_one_in=1000000 --periodic_compaction_seconds=1000 --prefix_size=5 --prefixpercent=5 --prepopulate_block_cache=1 --preserve_internal_time_seconds=0 --progress_reports=0 --promote_l0_one_in=0 --read_amp_bytes_per_bit=0 --read_fault_one_in=1000 --readahead_size=16384 --readpercent=45 --recycle_log_file_num=0 --reopen=0 --report_bg_io_stats=0 --reset_stats_one_in=1000000 --sample_for_compression=5 --secondary_cache_fault_one_in=32 --secondary_cache_uri= --set_options_one_in=0 --skip_stats_update_on_db_open=0 --snapshot_hold_ops=100000 --soft_pending_compaction_bytes_limit=68719476736 --sst_file_manager_bytes_per_sec=104857600 --sst_file_manager_bytes_per_truncate=1048576 --stats_dump_period_sec=0 --stats_history_buffer_size=0 --strict_bytes_per_sync=0 --subcompactions=2 --sync=0 --sync_fault_injection=1 --table_cache_numshardbits=0 --target_file_size_base=16777216 --target_file_size_multiplier=1 --test_batches_snapshots=0 --top_level_index_pinning=0 --uncache_aggressiveness=1 --universal_max_read_amp=-1 --unpartitioned_pinning=0 --use_adaptive_mutex=0 --use_adaptive_mutex_lru=0 --use_attribute_group=1 --use_delta_encoding=1 --use_direct_io_for_flush_and_compaction=1 --use_direct_reads=0 --use_full_merge_v1=0 --use_get_entity=0 --use_merge=0 --use_multi_cf_iterator=0 --use_multi_get_entity=1 --use_multiget=0 --use_put_entity_one_in=5 --use_timed_put_one_in=0 --use_write_buffer_manager=0 --user_timestamp_size=0 --value_size_mult=32 --verification_only=0 --verify_checksum=1 --verify_checksum_one_in=10 --verify_compression=1 --verify_db_one_in=10000 --verify_file_checksums_one_in=10 --verify_iterator_with_expected_state_one_in=5 --verify_sst_unique_id_in_manifest=1 --wal_bytes_per_sync=0 --wal_compression=zstd --write_buffer_size=33554432 --write_dbid_to_manifest=0 --write_fault_one_in=0 --writepercent=35 Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` - Run python3 tools/db_crashtest.py --simple blackbox --lock_wal_one_in=10 --backup_one_in=10 --sync_fault_injection=0 --use_direct_io_for_flush_and_compaction=0 for 1 hour - Monitor stress test CI Reviewed By: pdillinger Differential Revision: D58395807 Pulled By: hx235 fbshipit-source-id: 7d4b321acc0a0af3501b62dc417a7f6e2d318265
2024-06-18 21:41:14 +00:00
FSFileState() : pos_at_last_append_(-1), pos_at_last_sync_(-1) {}
Enable reading un-synced data in db stress test (#12752) Summary: **Context/Summary:** There are a few blockers to enabling reading un-synced data in db stress test (1) GetFileSize() will always return 0 for file written under direct IO because we don't track the last flushed position for `TestFSWritableFile` under direct IO. So it will surface as ``` Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` (2) A couple minor FIXME in left in https://github.com/facebook/rocksdb/pull/12729. This PR fixed (1) and (2) and enabled reading un-synced data in stress test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12752 Test Plan: - The following command failed before this PR and passed after. ``` ./db_stress --WAL_size_limit_MB=1 --WAL_ttl_seconds=0 --acquire_snapshot_one_in=100 --adaptive_readahead=0 --adm_policy=0 --advise_random_on_open=1 --allow_concurrent_memtable_write=0 --allow_data_in_errors=True --allow_fallocate=1 --async_io=1 --atomic_flush=1 --auto_readahead_size=1 --avoid_flush_during_recovery=0 --avoid_flush_during_shutdown=1 --avoid_unnecessary_blocking_io=0 --backup_max_size=104857600 --backup_one_in=100000 --batch_protection_bytes_per_key=0 --bgerror_resume_retry_interval=10000 --block_align=0 --block_protection_bytes_per_key=4 --block_size=16384 --bloom_before_level=2147483647 --bloom_bits=37.92024930098943 --bottommost_compression_type=disable --bottommost_file_compaction_delay=0 --bytes_per_sync=0 --cache_index_and_filter_blocks=1 --cache_index_and_filter_blocks_with_high_priority=0 --cache_size=8388608 --cache_type=auto_hyper_clock_cache --charge_compression_dictionary_building_buffer=1 --charge_file_metadata=0 --charge_filter_construction=0 --charge_table_reader=1 --check_multiget_consistency=0 --check_multiget_entity_consistency=0 --checkpoint_one_in=1000000 --checksum_type=kXXH3 --clear_column_family_one_in=0 --column_families=1 --compact_files_one_in=1000 --compact_range_one_in=1000 --compaction_pri=3 --compaction_readahead_size=0 --compaction_ttl=10 --compress_format_version=2 --compressed_secondary_cache_size=8388608 --compression_checksum=1 --compression_max_dict_buffer_bytes=0 --compression_max_dict_bytes=0 --compression_parallel_threads=1 --compression_type=zlib --compression_use_zstd_dict_trainer=0 --compression_zstd_max_train_bytes=0 --continuous_verification_interval=0 --daily_offpeak_time_utc= --data_block_index_type=0 --db=/tmp/rocksdb_crashtest_blackbox4deg_c5e --db_write_buffer_size=0 --default_temperature=kWarm --default_write_temperature=kHot --delete_obsolete_files_period_micros=30000000 --delpercent=4 --delrangepercent=1 --destroy_db_initially=0 --detect_filter_construct_corruption=0 --disable_file_deletions_one_in=1000000 --disable_manual_compaction_one_in=10000 --disable_wal=1 --dump_malloc_stats=0 --enable_checksum_handoff=0 --enable_compaction_filter=0 --enable_custom_split_merge=1 --enable_do_not_compress_roles=0 --enable_index_compression=1 --enable_memtable_insert_with_hint_prefix_extractor=0 --enable_pipelined_write=0 --enable_sst_partitioner_factory=1 --enable_thread_tracking=0 --enable_write_thread_adaptive_yield=0 --expected_values_dir=/tmp/rocksdb_crashtest_expected_8whyhdxm --fail_if_options_file_error=0 --fifo_allow_compaction=1 --file_checksum_impl=xxh64 --fill_cache=1 --flush_one_in=1000 --format_version=4 --get_all_column_family_metadata_one_in=10000 --get_current_wal_file_one_in=0 --get_live_files_apis_one_in=1000000 --get_properties_of_all_tables_one_in=100000 --get_property_one_in=100000 --get_sorted_wal_files_one_in=0 --hard_pending_compaction_bytes_limit=274877906944 --high_pri_pool_ratio=0.5 --index_block_restart_interval=9 --index_shortening=0 --index_type=0 --ingest_external_file_one_in=0 --initial_auto_readahead_size=0 --inplace_update_support=0 --iterpercent=10 --key_len_percent_dist=1,30,69 --key_may_exist_one_in=100000 --last_level_temperature=kUnknown --level_compaction_dynamic_level_bytes=1 --lock_wal_one_in=100 --log_file_time_to_roll=0 --log_readahead_size=0 --long_running_snapshots=0 --low_pri_pool_ratio=0 --lowest_used_cache_tier=0 --manifest_preallocation_size=5120 --manual_wal_flush_one_in=0 --mark_for_compaction_one_file_in=0 --max_auto_readahead_size=0 --max_background_compactions=1 --max_bytes_for_level_base=67108864 --max_key=100000 --max_key_len=3 --max_log_file_size=0 --max_manifest_file_size=1073741824 --max_sequential_skip_in_iterations=2 --max_total_wal_size=0 --max_write_batch_group_size_bytes=1048576 --max_write_buffer_number=3 --max_write_buffer_size_to_maintain=2097152 --memtable_insert_hint_per_batch=1 --memtable_max_range_deletions=100 --memtable_prefix_bloom_size_ratio=0.001 --memtable_protection_bytes_per_key=0 --memtable_whole_key_filtering=0 --memtablerep=skip_list --metadata_charge_policy=1 --min_write_buffer_number_to_merge=2 --mmap_read=0 --mock_direct_io=False --nooverwritepercent=1 --num_file_reads_for_auto_readahead=1 --open_files=-1 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=16 --ops_per_thread=100000000 --optimize_filters_for_hits=1 --optimize_filters_for_memory=0 --optimize_multiget_for_io=0 --paranoid_file_checks=1 --partition_filters=0 --partition_pinning=1 --pause_background_one_in=1000000 --periodic_compaction_seconds=1000 --prefix_size=5 --prefixpercent=5 --prepopulate_block_cache=1 --preserve_internal_time_seconds=0 --progress_reports=0 --promote_l0_one_in=0 --read_amp_bytes_per_bit=0 --read_fault_one_in=1000 --readahead_size=16384 --readpercent=45 --recycle_log_file_num=0 --reopen=0 --report_bg_io_stats=0 --reset_stats_one_in=1000000 --sample_for_compression=5 --secondary_cache_fault_one_in=32 --secondary_cache_uri= --set_options_one_in=0 --skip_stats_update_on_db_open=0 --snapshot_hold_ops=100000 --soft_pending_compaction_bytes_limit=68719476736 --sst_file_manager_bytes_per_sec=104857600 --sst_file_manager_bytes_per_truncate=1048576 --stats_dump_period_sec=0 --stats_history_buffer_size=0 --strict_bytes_per_sync=0 --subcompactions=2 --sync=0 --sync_fault_injection=1 --table_cache_numshardbits=0 --target_file_size_base=16777216 --target_file_size_multiplier=1 --test_batches_snapshots=0 --top_level_index_pinning=0 --uncache_aggressiveness=1 --universal_max_read_amp=-1 --unpartitioned_pinning=0 --use_adaptive_mutex=0 --use_adaptive_mutex_lru=0 --use_attribute_group=1 --use_delta_encoding=1 --use_direct_io_for_flush_and_compaction=1 --use_direct_reads=0 --use_full_merge_v1=0 --use_get_entity=0 --use_merge=0 --use_multi_cf_iterator=0 --use_multi_get_entity=1 --use_multiget=0 --use_put_entity_one_in=5 --use_timed_put_one_in=0 --use_write_buffer_manager=0 --user_timestamp_size=0 --value_size_mult=32 --verification_only=0 --verify_checksum=1 --verify_checksum_one_in=10 --verify_compression=1 --verify_db_one_in=10000 --verify_file_checksums_one_in=10 --verify_iterator_with_expected_state_one_in=5 --verify_sst_unique_id_in_manifest=1 --wal_bytes_per_sync=0 --wal_compression=zstd --write_buffer_size=33554432 --write_dbid_to_manifest=0 --write_fault_one_in=0 --writepercent=35 Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` - Run python3 tools/db_crashtest.py --simple blackbox --lock_wal_one_in=10 --backup_one_in=10 --sync_fault_injection=0 --use_direct_io_for_flush_and_compaction=0 for 1 hour - Monitor stress test CI Reviewed By: pdillinger Differential Revision: D58395807 Pulled By: hx235 fbshipit-source-id: 7d4b321acc0a0af3501b62dc417a7f6e2d318265
2024-06-18 21:41:14 +00:00
bool IsFullySynced() const {
return pos_at_last_append_ <= 0 || pos_at_last_append_ == pos_at_last_sync_;
}
IOStatus DropUnsyncedData();
IOStatus DropRandomUnsyncedData(Random* rand);
};
// A wrapper around WritableFileWriter* file
// is written to or sync'ed.
class TestFSWritableFile : public FSWritableFile {
public:
explicit TestFSWritableFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>&& f,
FaultInjectionTestFS* fs);
virtual ~TestFSWritableFile();
IOStatus Append(const Slice& data, const IOOptions&,
IODebugContext*) override;
IOStatus Append(const Slice& data, const IOOptions& options,
const DataVerificationInfo& verification_info,
IODebugContext* dbg) override;
IOStatus Truncate(uint64_t size, const IOOptions& options,
Enable reading un-synced data in db stress test (#12752) Summary: **Context/Summary:** There are a few blockers to enabling reading un-synced data in db stress test (1) GetFileSize() will always return 0 for file written under direct IO because we don't track the last flushed position for `TestFSWritableFile` under direct IO. So it will surface as ``` Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` (2) A couple minor FIXME in left in https://github.com/facebook/rocksdb/pull/12729. This PR fixed (1) and (2) and enabled reading un-synced data in stress test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12752 Test Plan: - The following command failed before this PR and passed after. ``` ./db_stress --WAL_size_limit_MB=1 --WAL_ttl_seconds=0 --acquire_snapshot_one_in=100 --adaptive_readahead=0 --adm_policy=0 --advise_random_on_open=1 --allow_concurrent_memtable_write=0 --allow_data_in_errors=True --allow_fallocate=1 --async_io=1 --atomic_flush=1 --auto_readahead_size=1 --avoid_flush_during_recovery=0 --avoid_flush_during_shutdown=1 --avoid_unnecessary_blocking_io=0 --backup_max_size=104857600 --backup_one_in=100000 --batch_protection_bytes_per_key=0 --bgerror_resume_retry_interval=10000 --block_align=0 --block_protection_bytes_per_key=4 --block_size=16384 --bloom_before_level=2147483647 --bloom_bits=37.92024930098943 --bottommost_compression_type=disable --bottommost_file_compaction_delay=0 --bytes_per_sync=0 --cache_index_and_filter_blocks=1 --cache_index_and_filter_blocks_with_high_priority=0 --cache_size=8388608 --cache_type=auto_hyper_clock_cache --charge_compression_dictionary_building_buffer=1 --charge_file_metadata=0 --charge_filter_construction=0 --charge_table_reader=1 --check_multiget_consistency=0 --check_multiget_entity_consistency=0 --checkpoint_one_in=1000000 --checksum_type=kXXH3 --clear_column_family_one_in=0 --column_families=1 --compact_files_one_in=1000 --compact_range_one_in=1000 --compaction_pri=3 --compaction_readahead_size=0 --compaction_ttl=10 --compress_format_version=2 --compressed_secondary_cache_size=8388608 --compression_checksum=1 --compression_max_dict_buffer_bytes=0 --compression_max_dict_bytes=0 --compression_parallel_threads=1 --compression_type=zlib --compression_use_zstd_dict_trainer=0 --compression_zstd_max_train_bytes=0 --continuous_verification_interval=0 --daily_offpeak_time_utc= --data_block_index_type=0 --db=/tmp/rocksdb_crashtest_blackbox4deg_c5e --db_write_buffer_size=0 --default_temperature=kWarm --default_write_temperature=kHot --delete_obsolete_files_period_micros=30000000 --delpercent=4 --delrangepercent=1 --destroy_db_initially=0 --detect_filter_construct_corruption=0 --disable_file_deletions_one_in=1000000 --disable_manual_compaction_one_in=10000 --disable_wal=1 --dump_malloc_stats=0 --enable_checksum_handoff=0 --enable_compaction_filter=0 --enable_custom_split_merge=1 --enable_do_not_compress_roles=0 --enable_index_compression=1 --enable_memtable_insert_with_hint_prefix_extractor=0 --enable_pipelined_write=0 --enable_sst_partitioner_factory=1 --enable_thread_tracking=0 --enable_write_thread_adaptive_yield=0 --expected_values_dir=/tmp/rocksdb_crashtest_expected_8whyhdxm --fail_if_options_file_error=0 --fifo_allow_compaction=1 --file_checksum_impl=xxh64 --fill_cache=1 --flush_one_in=1000 --format_version=4 --get_all_column_family_metadata_one_in=10000 --get_current_wal_file_one_in=0 --get_live_files_apis_one_in=1000000 --get_properties_of_all_tables_one_in=100000 --get_property_one_in=100000 --get_sorted_wal_files_one_in=0 --hard_pending_compaction_bytes_limit=274877906944 --high_pri_pool_ratio=0.5 --index_block_restart_interval=9 --index_shortening=0 --index_type=0 --ingest_external_file_one_in=0 --initial_auto_readahead_size=0 --inplace_update_support=0 --iterpercent=10 --key_len_percent_dist=1,30,69 --key_may_exist_one_in=100000 --last_level_temperature=kUnknown --level_compaction_dynamic_level_bytes=1 --lock_wal_one_in=100 --log_file_time_to_roll=0 --log_readahead_size=0 --long_running_snapshots=0 --low_pri_pool_ratio=0 --lowest_used_cache_tier=0 --manifest_preallocation_size=5120 --manual_wal_flush_one_in=0 --mark_for_compaction_one_file_in=0 --max_auto_readahead_size=0 --max_background_compactions=1 --max_bytes_for_level_base=67108864 --max_key=100000 --max_key_len=3 --max_log_file_size=0 --max_manifest_file_size=1073741824 --max_sequential_skip_in_iterations=2 --max_total_wal_size=0 --max_write_batch_group_size_bytes=1048576 --max_write_buffer_number=3 --max_write_buffer_size_to_maintain=2097152 --memtable_insert_hint_per_batch=1 --memtable_max_range_deletions=100 --memtable_prefix_bloom_size_ratio=0.001 --memtable_protection_bytes_per_key=0 --memtable_whole_key_filtering=0 --memtablerep=skip_list --metadata_charge_policy=1 --min_write_buffer_number_to_merge=2 --mmap_read=0 --mock_direct_io=False --nooverwritepercent=1 --num_file_reads_for_auto_readahead=1 --open_files=-1 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=16 --ops_per_thread=100000000 --optimize_filters_for_hits=1 --optimize_filters_for_memory=0 --optimize_multiget_for_io=0 --paranoid_file_checks=1 --partition_filters=0 --partition_pinning=1 --pause_background_one_in=1000000 --periodic_compaction_seconds=1000 --prefix_size=5 --prefixpercent=5 --prepopulate_block_cache=1 --preserve_internal_time_seconds=0 --progress_reports=0 --promote_l0_one_in=0 --read_amp_bytes_per_bit=0 --read_fault_one_in=1000 --readahead_size=16384 --readpercent=45 --recycle_log_file_num=0 --reopen=0 --report_bg_io_stats=0 --reset_stats_one_in=1000000 --sample_for_compression=5 --secondary_cache_fault_one_in=32 --secondary_cache_uri= --set_options_one_in=0 --skip_stats_update_on_db_open=0 --snapshot_hold_ops=100000 --soft_pending_compaction_bytes_limit=68719476736 --sst_file_manager_bytes_per_sec=104857600 --sst_file_manager_bytes_per_truncate=1048576 --stats_dump_period_sec=0 --stats_history_buffer_size=0 --strict_bytes_per_sync=0 --subcompactions=2 --sync=0 --sync_fault_injection=1 --table_cache_numshardbits=0 --target_file_size_base=16777216 --target_file_size_multiplier=1 --test_batches_snapshots=0 --top_level_index_pinning=0 --uncache_aggressiveness=1 --universal_max_read_amp=-1 --unpartitioned_pinning=0 --use_adaptive_mutex=0 --use_adaptive_mutex_lru=0 --use_attribute_group=1 --use_delta_encoding=1 --use_direct_io_for_flush_and_compaction=1 --use_direct_reads=0 --use_full_merge_v1=0 --use_get_entity=0 --use_merge=0 --use_multi_cf_iterator=0 --use_multi_get_entity=1 --use_multiget=0 --use_put_entity_one_in=5 --use_timed_put_one_in=0 --use_write_buffer_manager=0 --user_timestamp_size=0 --value_size_mult=32 --verification_only=0 --verify_checksum=1 --verify_checksum_one_in=10 --verify_compression=1 --verify_db_one_in=10000 --verify_file_checksums_one_in=10 --verify_iterator_with_expected_state_one_in=5 --verify_sst_unique_id_in_manifest=1 --wal_bytes_per_sync=0 --wal_compression=zstd --write_buffer_size=33554432 --write_dbid_to_manifest=0 --write_fault_one_in=0 --writepercent=35 Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` - Run python3 tools/db_crashtest.py --simple blackbox --lock_wal_one_in=10 --backup_one_in=10 --sync_fault_injection=0 --use_direct_io_for_flush_and_compaction=0 for 1 hour - Monitor stress test CI Reviewed By: pdillinger Differential Revision: D58395807 Pulled By: hx235 fbshipit-source-id: 7d4b321acc0a0af3501b62dc417a7f6e2d318265
2024-06-18 21:41:14 +00:00
IODebugContext* dbg) override;
IOStatus Close(const IOOptions& options, IODebugContext* dbg) override;
IOStatus Flush(const IOOptions&, IODebugContext*) override;
IOStatus Sync(const IOOptions& options, IODebugContext* dbg) override;
IOStatus RangeSync(uint64_t /*offset*/, uint64_t /*nbytes*/,
const IOOptions& options, IODebugContext* dbg) override;
bool IsSyncThreadSafe() const override { return true; }
IOStatus PositionedAppend(const Slice& data, uint64_t offset,
const IOOptions& options,
Enable reading un-synced data in db stress test (#12752) Summary: **Context/Summary:** There are a few blockers to enabling reading un-synced data in db stress test (1) GetFileSize() will always return 0 for file written under direct IO because we don't track the last flushed position for `TestFSWritableFile` under direct IO. So it will surface as ``` Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` (2) A couple minor FIXME in left in https://github.com/facebook/rocksdb/pull/12729. This PR fixed (1) and (2) and enabled reading un-synced data in stress test. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12752 Test Plan: - The following command failed before this PR and passed after. ``` ./db_stress --WAL_size_limit_MB=1 --WAL_ttl_seconds=0 --acquire_snapshot_one_in=100 --adaptive_readahead=0 --adm_policy=0 --advise_random_on_open=1 --allow_concurrent_memtable_write=0 --allow_data_in_errors=True --allow_fallocate=1 --async_io=1 --atomic_flush=1 --auto_readahead_size=1 --avoid_flush_during_recovery=0 --avoid_flush_during_shutdown=1 --avoid_unnecessary_blocking_io=0 --backup_max_size=104857600 --backup_one_in=100000 --batch_protection_bytes_per_key=0 --bgerror_resume_retry_interval=10000 --block_align=0 --block_protection_bytes_per_key=4 --block_size=16384 --bloom_before_level=2147483647 --bloom_bits=37.92024930098943 --bottommost_compression_type=disable --bottommost_file_compaction_delay=0 --bytes_per_sync=0 --cache_index_and_filter_blocks=1 --cache_index_and_filter_blocks_with_high_priority=0 --cache_size=8388608 --cache_type=auto_hyper_clock_cache --charge_compression_dictionary_building_buffer=1 --charge_file_metadata=0 --charge_filter_construction=0 --charge_table_reader=1 --check_multiget_consistency=0 --check_multiget_entity_consistency=0 --checkpoint_one_in=1000000 --checksum_type=kXXH3 --clear_column_family_one_in=0 --column_families=1 --compact_files_one_in=1000 --compact_range_one_in=1000 --compaction_pri=3 --compaction_readahead_size=0 --compaction_ttl=10 --compress_format_version=2 --compressed_secondary_cache_size=8388608 --compression_checksum=1 --compression_max_dict_buffer_bytes=0 --compression_max_dict_bytes=0 --compression_parallel_threads=1 --compression_type=zlib --compression_use_zstd_dict_trainer=0 --compression_zstd_max_train_bytes=0 --continuous_verification_interval=0 --daily_offpeak_time_utc= --data_block_index_type=0 --db=/tmp/rocksdb_crashtest_blackbox4deg_c5e --db_write_buffer_size=0 --default_temperature=kWarm --default_write_temperature=kHot --delete_obsolete_files_period_micros=30000000 --delpercent=4 --delrangepercent=1 --destroy_db_initially=0 --detect_filter_construct_corruption=0 --disable_file_deletions_one_in=1000000 --disable_manual_compaction_one_in=10000 --disable_wal=1 --dump_malloc_stats=0 --enable_checksum_handoff=0 --enable_compaction_filter=0 --enable_custom_split_merge=1 --enable_do_not_compress_roles=0 --enable_index_compression=1 --enable_memtable_insert_with_hint_prefix_extractor=0 --enable_pipelined_write=0 --enable_sst_partitioner_factory=1 --enable_thread_tracking=0 --enable_write_thread_adaptive_yield=0 --expected_values_dir=/tmp/rocksdb_crashtest_expected_8whyhdxm --fail_if_options_file_error=0 --fifo_allow_compaction=1 --file_checksum_impl=xxh64 --fill_cache=1 --flush_one_in=1000 --format_version=4 --get_all_column_family_metadata_one_in=10000 --get_current_wal_file_one_in=0 --get_live_files_apis_one_in=1000000 --get_properties_of_all_tables_one_in=100000 --get_property_one_in=100000 --get_sorted_wal_files_one_in=0 --hard_pending_compaction_bytes_limit=274877906944 --high_pri_pool_ratio=0.5 --index_block_restart_interval=9 --index_shortening=0 --index_type=0 --ingest_external_file_one_in=0 --initial_auto_readahead_size=0 --inplace_update_support=0 --iterpercent=10 --key_len_percent_dist=1,30,69 --key_may_exist_one_in=100000 --last_level_temperature=kUnknown --level_compaction_dynamic_level_bytes=1 --lock_wal_one_in=100 --log_file_time_to_roll=0 --log_readahead_size=0 --long_running_snapshots=0 --low_pri_pool_ratio=0 --lowest_used_cache_tier=0 --manifest_preallocation_size=5120 --manual_wal_flush_one_in=0 --mark_for_compaction_one_file_in=0 --max_auto_readahead_size=0 --max_background_compactions=1 --max_bytes_for_level_base=67108864 --max_key=100000 --max_key_len=3 --max_log_file_size=0 --max_manifest_file_size=1073741824 --max_sequential_skip_in_iterations=2 --max_total_wal_size=0 --max_write_batch_group_size_bytes=1048576 --max_write_buffer_number=3 --max_write_buffer_size_to_maintain=2097152 --memtable_insert_hint_per_batch=1 --memtable_max_range_deletions=100 --memtable_prefix_bloom_size_ratio=0.001 --memtable_protection_bytes_per_key=0 --memtable_whole_key_filtering=0 --memtablerep=skip_list --metadata_charge_policy=1 --min_write_buffer_number_to_merge=2 --mmap_read=0 --mock_direct_io=False --nooverwritepercent=1 --num_file_reads_for_auto_readahead=1 --open_files=-1 --open_metadata_write_fault_one_in=0 --open_read_fault_one_in=0 --open_write_fault_one_in=16 --ops_per_thread=100000000 --optimize_filters_for_hits=1 --optimize_filters_for_memory=0 --optimize_multiget_for_io=0 --paranoid_file_checks=1 --partition_filters=0 --partition_pinning=1 --pause_background_one_in=1000000 --periodic_compaction_seconds=1000 --prefix_size=5 --prefixpercent=5 --prepopulate_block_cache=1 --preserve_internal_time_seconds=0 --progress_reports=0 --promote_l0_one_in=0 --read_amp_bytes_per_bit=0 --read_fault_one_in=1000 --readahead_size=16384 --readpercent=45 --recycle_log_file_num=0 --reopen=0 --report_bg_io_stats=0 --reset_stats_one_in=1000000 --sample_for_compression=5 --secondary_cache_fault_one_in=32 --secondary_cache_uri= --set_options_one_in=0 --skip_stats_update_on_db_open=0 --snapshot_hold_ops=100000 --soft_pending_compaction_bytes_limit=68719476736 --sst_file_manager_bytes_per_sec=104857600 --sst_file_manager_bytes_per_truncate=1048576 --stats_dump_period_sec=0 --stats_history_buffer_size=0 --strict_bytes_per_sync=0 --subcompactions=2 --sync=0 --sync_fault_injection=1 --table_cache_numshardbits=0 --target_file_size_base=16777216 --target_file_size_multiplier=1 --test_batches_snapshots=0 --top_level_index_pinning=0 --uncache_aggressiveness=1 --universal_max_read_amp=-1 --unpartitioned_pinning=0 --use_adaptive_mutex=0 --use_adaptive_mutex_lru=0 --use_attribute_group=1 --use_delta_encoding=1 --use_direct_io_for_flush_and_compaction=1 --use_direct_reads=0 --use_full_merge_v1=0 --use_get_entity=0 --use_merge=0 --use_multi_cf_iterator=0 --use_multi_get_entity=1 --use_multiget=0 --use_put_entity_one_in=5 --use_timed_put_one_in=0 --use_write_buffer_manager=0 --user_timestamp_size=0 --value_size_mult=32 --verification_only=0 --verify_checksum=1 --verify_checksum_one_in=10 --verify_compression=1 --verify_db_one_in=10000 --verify_file_checksums_one_in=10 --verify_iterator_with_expected_state_one_in=5 --verify_sst_unique_id_in_manifest=1 --wal_bytes_per_sync=0 --wal_compression=zstd --write_buffer_size=33554432 --write_dbid_to_manifest=0 --write_fault_one_in=0 --writepercent=35 Verification failed: VerifyChecksum failed: Corruption: file is too short (0 bytes) to be an sstable: /tmp/rocksdb_crashtest_blackbox4deg_c5e/000009.sst db_stress: db_stress_tool/db_stress_test_base.cc:518: void rocksdb::StressTest::ProcessStatus(rocksdb::SharedState*, std::string, const rocksdb::Status&, bool) const: Assertion `false' failed. Received signal 6 (Aborted) Invoking GDB for stack trace... ``` - Run python3 tools/db_crashtest.py --simple blackbox --lock_wal_one_in=10 --backup_one_in=10 --sync_fault_injection=0 --use_direct_io_for_flush_and_compaction=0 for 1 hour - Monitor stress test CI Reviewed By: pdillinger Differential Revision: D58395807 Pulled By: hx235 fbshipit-source-id: 7d4b321acc0a0af3501b62dc417a7f6e2d318265
2024-06-18 21:41:14 +00:00
IODebugContext* dbg) override;
IOStatus PositionedAppend(const Slice& data, uint64_t offset,
const IOOptions& options,
Using existing crc32c checksum in checksum handoff for Manifest and WAL (#8412) Summary: In PR https://github.com/facebook/rocksdb/issues/7523 , checksum handoff is introduced in RocksDB for WAL, Manifest, and SST files. When user enable checksum handoff for a certain type of file, before the data is written to the lower layer storage system, we calculate the checksum (crc32c) of each piece of data and pass the checksum down with the data, such that data verification can be down by the lower layer storage system if it has the capability. However, it cannot cover the whole lifetime of the data in the memory and also it potentially introduces extra checksum calculation overhead. In this PR, we introduce a new interface in WritableFileWriter::Append, which allows the caller be able to pass the data and the checksum (crc32c) together. In this way, WritableFileWriter can directly use the pass-in checksum (crc32c) to generate the checksum of data being passed down to the storage system. It saves the calculation overhead and achieves higher protection coverage. When a new checksum is added with the data, we use Crc32cCombine https://github.com/facebook/rocksdb/issues/8305 to combine the existing checksum and the new checksum. To avoid the segmenting of data by rate-limiter before it is stored, rate-limiter is called enough times to accumulate enough credits for a certain write. This design only support Manifest and WAL which use log_writer in the current stage. Pull Request resolved: https://github.com/facebook/rocksdb/pull/8412 Test Plan: make check, add new testing cases. Reviewed By: anand1976 Differential Revision: D29151545 Pulled By: zhichao-cao fbshipit-source-id: 75e2278c5126cfd58393c67b1efd18dcc7a30772
2021-06-25 07:46:33 +00:00
const DataVerificationInfo& verification_info,
IODebugContext* dbg) override;
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
bool use_direct_io() const override { return target_->use_direct_io(); }
uint64_t GetFileSize(const IOOptions& options, IODebugContext* dbg) override {
MutexLock l(&mutex_);
return target_->GetFileSize(options, dbg);
}
private:
Add missing synchronization in TestFSWritableFile (#10544) Summary: **Context:** ajkr's command revealed an existing TSAN data race between `TestFSWritableFile::Append` and `TestFSWritableFile::Sync` on `TestFSWritableFile::state_` ``` $ make clean && COMPILE_WITH_TSAN=1 make -j56 db_stress $ python3 tools/db_crashtest.py blackbox --simple --duration=3600 --interval=10 --sync_fault_injection=1 --disable_wal=0 --max_key=10000 --checkpoint_one_in=1000 ``` The race is due to concurrent access from [checkpoint's WAL sync](https://github.com/facebook/rocksdb/blob/7.4.fb/utilities/fault_injection_fs.cc#L324) and [db put's WAL write when ‘sync_fault_injection=1 ‘](https://github.com/facebook/rocksdb/blob/7.4.fb/utilities/fault_injection_fs.cc#L208) to the `state_` on the same WAL `TestFSWritableFile` under the missing synchronization. ``` WARNING: ThreadSanitizer: data race (pid=11275) Write of size 8 at 0x7b480003d850 by thread T23 (mutexes: write M69230): #0 rocksdb::TestFSWritableFile::Sync(rocksdb::IOOptions const&, rocksdb::IODebugContext*) internal_repo_rocksdb/repo/utilities/fault_injection_fs.cc:297 (db_stress+0x716004) https://github.com/facebook/rocksdb/issues/1 rocksdb::(anonymous namespace)::CompositeWritableFileWrapper::Sync() internal_repo_rocksdb/repo/env/composite_env.cc:154 (db_stress+0x4dfa78) https://github.com/facebook/rocksdb/issues/2 rocksdb::(anonymous namespace)::LegacyWritableFileWrapper::Sync(rocksdb::IOOptions const&, rocksdb::IODebugContext*) internal_repo_rocksdb/repo/env/env.cc:280 (db_stress+0x6dfd24) https://github.com/facebook/rocksdb/issues/3 rocksdb::WritableFileWriter::SyncInternal(bool) internal_repo_rocksdb/repo/file/writable_file_writer.cc:460 (db_stress+0xa1b98c) https://github.com/facebook/rocksdb/issues/4 rocksdb::WritableFileWriter::SyncWithoutFlush(bool) internal_repo_rocksdb/repo/file/writable_file_writer.cc:435 (db_stress+0xa1e441) https://github.com/facebook/rocksdb/issues/5 rocksdb::DBImpl::SyncWAL() internal_repo_rocksdb/repo/db/db_impl/db_impl.cc:1385 (db_stress+0x529458) https://github.com/facebook/rocksdb/issues/6 rocksdb::DBImpl::FlushWAL(bool) internal_repo_rocksdb/repo/db/db_impl/db_impl.cc:1339 (db_stress+0x54f82a) https://github.com/facebook/rocksdb/issues/7 rocksdb::DBImpl::GetLiveFilesStorageInfo(rocksdb::LiveFilesStorageInfoOptions const&, std::vector<rocksdb::LiveFileStorageInfo, std::allocator<rocksdb::LiveFileStorageInfo> >*) internal_repo_rocksdb/repo/db/db_filesnapshot.cc:387 (db_stress+0x5c831d) https://github.com/facebook/rocksdb/issues/8 rocksdb::CheckpointImpl::CreateCustomCheckpoint(std::function<rocksdb::Status (std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, rocksdb::FileType)>, std::function<rocksdb::Status (std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, unsigned long, rocksdb::FileType, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, rocksdb::Temperature)>, std::function<rocksdb::Status (std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, rocksdb::FileType)>, unsigned long*, unsigned long, bool) internal_repo_rocksdb/repo/utilities/checkpoint/checkpoint_impl.cc:214 (db_stress+0x4c0343) https://github.com/facebook/rocksdb/issues/9 rocksdb::CheckpointImpl::CreateCheckpoint(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, unsigned long, unsigned long*) internal_repo_rocksdb/repo/utilities/checkpoint/checkpoint_impl.cc:123 (db_stress+0x4c237e) https://github.com/facebook/rocksdb/issues/10 rocksdb::StressTest::TestCheckpoint(rocksdb::ThreadState*, std::vector<int, std::allocator<int> > const&, std::vector<long, std::allocator<long> > const&) internal_repo_rocksdb/repo/db_stress_tool/db_stress_test_base.cc:1699 (db_stress+0x328340) https://github.com/facebook/rocksdb/issues/11 rocksdb::StressTest::OperateDb(rocksdb::ThreadState*) internal_repo_rocksdb/repo/db_stress_tool/db_stress_test_base.cc:825 (db_stress+0x33921f) https://github.com/facebook/rocksdb/issues/12 rocksdb::ThreadBody(void*) internal_repo_rocksdb/repo/db_stress_tool/db_stress_driver.cc:33 (db_stress+0x354857) https://github.com/facebook/rocksdb/issues/13 rocksdb::(anonymous namespace)::StartThreadWrapper(void*) internal_repo_rocksdb/repo/env/env_posix.cc:447 (db_stress+0x6eb2ad) Previous read of size 8 at 0x7b480003d850 by thread T64 (mutexes: write M980798978697532600, write M253744503184415024, write M1262): #0 memcpy <null> (db_stress+0xbc9696) https://github.com/facebook/rocksdb/issues/1 operator= internal_repo_rocksdb/repo/utilities/fault_injection_fs.h:35 (db_stress+0x70d5f1) https://github.com/facebook/rocksdb/issues/2 rocksdb::FaultInjectionTestFS::WritableFileAppended(rocksdb::FSFileState const&) internal_repo_rocksdb/repo/utilities/fault_injection_fs.cc:827 (db_stress+0x70d5f1) https://github.com/facebook/rocksdb/issues/3 rocksdb::TestFSWritableFile::Append(rocksdb::Slice const&, rocksdb::IOOptions const&, rocksdb::IODebugContext*) internal_repo_rocksdb/repo/utilities/fault_injection_fs.cc:173 (db_stress+0x7143af) https://github.com/facebook/rocksdb/issues/4 rocksdb::(anonymous namespace)::CompositeWritableFileWrapper::Append(rocksdb::Slice const&) internal_repo_rocksdb/repo/env/composite_env.cc:115 (db_stress+0x4de3ab) https://github.com/facebook/rocksdb/issues/5 rocksdb::(anonymous namespace)::LegacyWritableFileWrapper::Append(rocksdb::Slice const&, rocksdb::IOOptions const&, rocksdb::IODebugContext*) internal_repo_rocksdb/repo/env/env.cc:248 (db_stress+0x6df44b) https://github.com/facebook/rocksdb/issues/6 rocksdb::WritableFileWriter::WriteBuffered(char const*, unsigned long, rocksdb::Env::IOPriority) internal_repo_rocksdb/repo/file/writable_file_writer.cc:551 (db_stress+0xa1a953) https://github.com/facebook/rocksdb/issues/7 rocksdb::WritableFileWriter::Flush(rocksdb::Env::IOPriority) internal_repo_rocksdb/repo/file/writable_file_writer.cc:327 (db_stress+0xa16ee8) https://github.com/facebook/rocksdb/issues/8 rocksdb::log::Writer::AddRecord(rocksdb::Slice const&, rocksdb::Env::IOPriority) internal_repo_rocksdb/repo/db/log_writer.cc:147 (db_stress+0x7f121f) https://github.com/facebook/rocksdb/issues/9 rocksdb::DBImpl::WriteToWAL(rocksdb::WriteBatch const&, rocksdb::log::Writer*, unsigned long*, unsigned long*, rocksdb::Env::IOPriority, rocksdb::DBImpl::LogFileNumberSize&) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:1285 (db_stress+0x695042) https://github.com/facebook/rocksdb/issues/10 rocksdb::DBImpl::WriteToWAL(rocksdb::WriteThread::WriteGroup const&, rocksdb::log::Writer*, unsigned long*, bool, bool, unsigned long, rocksdb::DBImpl::LogFileNumberSize&) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:1328 (db_stress+0x6907e8) https://github.com/facebook/rocksdb/issues/11 rocksdb::DBImpl::PipelinedWriteImpl(rocksdb::WriteOptions const&, rocksdb::WriteBatch*, rocksdb::WriteCallback*, unsigned long*, unsigned long, bool, unsigned long*) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:731 (db_stress+0x68e8a7) https://github.com/facebook/rocksdb/issues/12 rocksdb::DBImpl::WriteImpl(rocksdb::WriteOptions const&, rocksdb::WriteBatch*, rocksdb::WriteCallback*, unsigned long*, unsigned long, bool, unsigned long*, unsigned long, rocksdb::PreReleaseCallback*, rocksdb::PostMemTableCallback*) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:283 (db_stress+0x688370) https://github.com/facebook/rocksdb/issues/13 rocksdb::DBImpl::Write(rocksdb::WriteOptions const&, rocksdb::WriteBatch*) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:126 (db_stress+0x69a7b5) https://github.com/facebook/rocksdb/issues/14 rocksdb::DB::Put(rocksdb::WriteOptions const&, rocksdb::ColumnFamilyHandle*, rocksdb::Slice const&, rocksdb::Slice const&, rocksdb::Slice const&) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:2247 (db_stress+0x698634) https://github.com/facebook/rocksdb/issues/15 rocksdb::DBImpl::Put(rocksdb::WriteOptions const&, rocksdb::ColumnFamilyHandle*, rocksdb::Slice const&, rocksdb::Slice const&, rocksdb::Slice const&) internal_repo_rocksdb/repo/db/db_impl/db_impl_write.cc:37 (db_stress+0x699868) https://github.com/facebook/rocksdb/issues/16 rocksdb::NonBatchedOpsStressTest::TestPut(rocksdb::ThreadState*, rocksdb::WriteOptions&, rocksdb::ReadOptions const&, std::vector<int, std::allocator<int> > const&, std::vector<long, std::allocator<long> > const&, char (&) [100], std::unique_ptr<rocksdb::MutexLock, std::default_delete<rocksdb::MutexLock> >&) internal_repo_rocksdb/repo/db_stress_tool/no_batched_ops_stress.cc:681 (db_stress+0x38d20c) https://github.com/facebook/rocksdb/issues/17 rocksdb::StressTest::OperateDb(rocksdb::ThreadState*) internal_repo_rocksdb/repo/db_stress_tool/db_stress_test_base.cc:897 (db_stress+0x3399ec) https://github.com/facebook/rocksdb/issues/18 rocksdb::ThreadBody(void*) internal_repo_rocksdb/repo/db_stress_tool/db_stress_driver.cc:33 (db_stress+0x354857) https://github.com/facebook/rocksdb/issues/19 rocksdb::(anonymous namespace)::StartThreadWrapper(void*) internal_repo_rocksdb/repo/env/env_posix.cc:447 (db_stress+0x6eb2ad) Location is heap block of size 352 at 0x7b480003d800 allocated by thread T23: #0 operator new(unsigned long) <null> (db_stress+0xb685dc) https://github.com/facebook/rocksdb/issues/1 rocksdb::FaultInjectionTestFS::NewWritableFile(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, rocksdb::FileOptions const&, std::unique_ptr<rocksdb::FSWritableFile, std::default_delete<rocksdb::FSWritableFile> >*, rocksdb::IODebugContext*) internal_repo_rocksdb/repo/utilities/fault_injection_fs.cc:506 (db_stress+0x711192) https://github.com/facebook/rocksdb/issues/2 rocksdb::CompositeEnv::NewWritableFile(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unique_ptr<rocksdb::WritableFile, std::default_delete<rocksdb::WritableFile> >*, rocksdb::EnvOptions const&) internal_repo_rocksdb/repo/env/composite_env.cc:329 (db_stress+0x4d33fa) https://github.com/facebook/rocksdb/issues/3 rocksdb::EnvWrapper::NewWritableFile(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unique_ptr<rocksdb::WritableFile, std::default_delete<rocksdb::WritableFile> >*, rocksdb::EnvOptions const&) internal_repo_rocksdb/repo/include/rocksdb/env.h:1425 (db_stress+0x300662) ... ``` **Summary:** - Added the missing lock in functions mentioned above along with three other functions with a similar need in TestFSWritableFile - Added clarification comment Pull Request resolved: https://github.com/facebook/rocksdb/pull/10544 Test Plan: - Past the above race condition repro Reviewed By: ajkr Differential Revision: D38886634 Pulled By: hx235 fbshipit-source-id: 0571bae9615f35b16fbd8168204607e306b1b486
2022-08-22 22:50:22 +00:00
FSFileState state_; // Need protection by mutex_
FileOptions file_opts_;
std::unique_ptr<FSWritableFile> target_;
bool writable_file_opened_;
FaultInjectionTestFS* fs_;
port::Mutex mutex_;
};
// A wrapper around WritableFileWriter* file
// is written to or sync'ed.
class TestFSRandomRWFile : public FSRandomRWFile {
public:
explicit TestFSRandomRWFile(const std::string& fname,
std::unique_ptr<FSRandomRWFile>&& f,
FaultInjectionTestFS* fs);
virtual ~TestFSRandomRWFile();
IOStatus Write(uint64_t offset, const Slice& data, const IOOptions& options,
IODebugContext* dbg) override;
IOStatus Read(uint64_t offset, size_t n, const IOOptions& options,
Slice* result, char* scratch,
IODebugContext* dbg) const override;
IOStatus Close(const IOOptions& options, IODebugContext* dbg) override;
IOStatus Flush(const IOOptions& options, IODebugContext* dbg) override;
IOStatus Sync(const IOOptions& options, IODebugContext* dbg) override;
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
bool use_direct_io() const override { return target_->use_direct_io(); }
private:
std::unique_ptr<FSRandomRWFile> target_;
bool file_opened_;
FaultInjectionTestFS* fs_;
};
class TestFSRandomAccessFile : public FSRandomAccessFile {
public:
explicit TestFSRandomAccessFile(const std::string& fname,
std::unique_ptr<FSRandomAccessFile>&& f,
FaultInjectionTestFS* fs);
~TestFSRandomAccessFile() override {}
IOStatus Read(uint64_t offset, size_t n, const IOOptions& options,
Slice* result, char* scratch,
IODebugContext* dbg) const override;
IOStatus ReadAsync(FSReadRequest& req, const IOOptions& opts,
std::function<void(FSReadRequest&, void*)> cb,
void* cb_arg, void** io_handle, IOHandleDeleter* del_fn,
IODebugContext* dbg) override;
IOStatus MultiRead(FSReadRequest* reqs, size_t num_reqs,
const IOOptions& options, IODebugContext* dbg) override;
size_t GetRequiredBufferAlignment() const override {
return target_->GetRequiredBufferAlignment();
}
bool use_direct_io() const override { return target_->use_direct_io(); }
size_t GetUniqueId(char* id, size_t max_size) const override;
private:
std::unique_ptr<FSRandomAccessFile> target_;
FaultInjectionTestFS* fs_;
};
class TestFSSequentialFile : public FSSequentialFileOwnerWrapper {
public:
explicit TestFSSequentialFile(std::unique_ptr<FSSequentialFile>&& f,
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
FaultInjectionTestFS* fs, std::string fname)
: FSSequentialFileOwnerWrapper(std::move(f)),
fs_(fs),
fname_(std::move(fname)) {}
IOStatus Read(size_t n, const IOOptions& options, Slice* result,
char* scratch, IODebugContext* dbg) override;
IOStatus PositionedRead(uint64_t offset, size_t n, const IOOptions& options,
Slice* result, char* scratch,
IODebugContext* dbg) override;
private:
FaultInjectionTestFS* fs_;
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
std::string fname_;
size_t read_pos_ = 0;
};
class TestFSDirectory : public FSDirectory {
public:
explicit TestFSDirectory(FaultInjectionTestFS* fs, std::string dirname,
FSDirectory* dir)
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
: fs_(fs), dirname_(std::move(dirname)), dir_(dir) {}
~TestFSDirectory() {}
IOStatus Fsync(const IOOptions& options, IODebugContext* dbg) override;
IOStatus Close(const IOOptions& options, IODebugContext* dbg) override;
IOStatus FsyncWithDirOptions(
const IOOptions& options, IODebugContext* dbg,
const DirFsyncOptions& dir_fsync_options) override;
private:
FaultInjectionTestFS* fs_;
std::string dirname_;
std::unique_ptr<FSDirectory> dir_;
};
class FaultInjectionTestFS : public FileSystemWrapper {
public:
explicit FaultInjectionTestFS(const std::shared_ptr<FileSystem>& base)
: FileSystemWrapper(base),
filesystem_active_(true),
filesystem_writable_(false),
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
read_unsynced_data_(true),
Ensure Close() before LinkFile() for WALs in Checkpoint (#12734) Summary: POSIX semantics for LinkFile (hard links) allow linking a file that is still being written two, with both the source and destination showing any subsequent writes to the source. This may not be practical semantics for some FileSystem implementations such as remote storage. They might only link the flushed or sync-ed file contents at time of LinkFile, or might even have undefined behavior if LinkFile is called on a file still open for write (not yet "sealed"). This change builds on https://github.com/facebook/rocksdb/issues/12731 to bring more hygiene to our handling of WAL files in Checkpoint. Specifically, we now Close WAL files as soon as they are either (a) inactive and fully synced, or (b) inactive and obsolete (so maybe never fully synced), rather than letting Close() happen in handling obsolete files (maybe a background thread). This should not be a performance issue as Close() should be trivial cost relative to other IO ops, but just in case: * We don't Close() while holding a mutex, to avoid blocking, and * The old behavior is available with a new kill switch option `background_close_inactive_wals`. Stacked on https://github.com/facebook/rocksdb/issues/12731 Pull Request resolved: https://github.com/facebook/rocksdb/pull/12734 Test Plan: Extended existing unit test, especially adding a hygiene check to FaultInjectionTestFS to detect LinkFile() on a file still open for writes. FaultInjectionTestFS already has relevant tracking data, and tests can opt out of the new check, as in a smoke test I have left for the old, deprecated functionality `background_close_inactive_wals=true`. Also ran lengthy blackbox_crash_test to ensure the hygiene check is OK with the crash test. (The only place I can find we use LinkFile in production is Checkpoint.) Reviewed By: cbi42 Differential Revision: D58295284 Pulled By: pdillinger fbshipit-source-id: 64d90ed8477e2366c19eaf9c4c5ad60b82cac5c6
2024-06-12 18:48:45 +00:00
allow_link_open_file_(false),
thread_local_error_(new ThreadLocalPtr(DeleteThreadLocalErrorContext)),
enable_write_error_injection_(false),
enable_metadata_write_error_injection_(false),
write_error_rand_(0),
write_error_one_in_(0),
metadata_write_error_one_in_(0),
read_error_one_in_(0),
ingest_data_corruption_before_write_(false),
checksum_handoff_func_type_(kCRC32c),
fail_get_file_unique_id_(false) {}
virtual ~FaultInjectionTestFS() { error_.PermitUncheckedError(); }
static const char* kClassName() { return "FaultInjectionTestFS"; }
const char* Name() const override { return kClassName(); }
IOStatus NewDirectory(const std::string& name, const IOOptions& options,
std::unique_ptr<FSDirectory>* result,
IODebugContext* dbg) override;
IOStatus NewWritableFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>* result,
IODebugContext* dbg) override;
IOStatus ReopenWritableFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>* result,
IODebugContext* dbg) override;
IOStatus ReuseWritableFile(const std::string& fname,
const std::string& old_fname,
const FileOptions& file_opts,
std::unique_ptr<FSWritableFile>* result,
IODebugContext* dbg) override;
IOStatus NewRandomRWFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSRandomRWFile>* result,
IODebugContext* dbg) override;
IOStatus NewRandomAccessFile(const std::string& fname,
const FileOptions& file_opts,
std::unique_ptr<FSRandomAccessFile>* result,
IODebugContext* dbg) override;
IOStatus NewSequentialFile(const std::string& f, const FileOptions& file_opts,
std::unique_ptr<FSSequentialFile>* r,
IODebugContext* dbg) override;
IOStatus DeleteFile(const std::string& f, const IOOptions& options,
IODebugContext* dbg) override;
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
IOStatus GetFileSize(const std::string& f, const IOOptions& options,
uint64_t* file_size, IODebugContext* dbg) override;
IOStatus RenameFile(const std::string& s, const std::string& t,
const IOOptions& options, IODebugContext* dbg) override;
IOStatus LinkFile(const std::string& src, const std::string& target,
const IOOptions& options, IODebugContext* dbg) override;
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
2021-10-11 23:22:10 +00:00
// Undef to eliminate clash on Windows
#undef GetFreeSpace
IOStatus GetFreeSpace(const std::string& path, const IOOptions& options,
uint64_t* disk_free, IODebugContext* dbg) override {
IOStatus io_s;
if (!IsFilesystemActive() &&
error_.subcode() == IOStatus::SubCode::kNoSpace) {
*disk_free = 0;
} else {
io_s = target()->GetFreeSpace(path, options, disk_free, dbg);
}
return io_s;
}
IOStatus Poll(std::vector<void*>& io_handles,
size_t min_completions) override;
IOStatus AbortIO(std::vector<void*>& io_handles) override;
void WritableFileClosed(const FSFileState& state);
void WritableFileSynced(const FSFileState& state);
void WritableFileAppended(const FSFileState& state);
IOStatus DropUnsyncedFileData();
IOStatus DropRandomUnsyncedFileData(Random* rnd);
IOStatus DeleteFilesCreatedAfterLastDirSync(const IOOptions& options,
IODebugContext* dbg);
void ResetState();
void UntrackFile(const std::string& f);
void SyncDir(const std::string& dirname) {
MutexLock l(&mutex_);
dir_to_new_files_since_last_sync_.erase(dirname);
}
// Setting the filesystem to inactive is the test equivalent to simulating a
// system reset. Setting to inactive will freeze our saved filesystem state so
// that it will stop being recorded. It can then be reset back to the state at
// the time of the reset.
bool IsFilesystemActive() {
MutexLock l(&mutex_);
return filesystem_active_;
}
// Setting filesystem_writable_ makes NewWritableFile. ReopenWritableFile,
// and NewRandomRWFile bypass FaultInjectionTestFS and go directly to the
// target FS
bool IsFilesystemDirectWritable() {
MutexLock l(&mutex_);
return filesystem_writable_;
}
bool ShouldUseDiretWritable(const std::string& file_name) {
MutexLock l(&mutex_);
if (filesystem_writable_) {
return true;
}
FileType file_type = kTempFile;
uint64_t file_number = 0;
if (!TryParseFileName(file_name, &file_number, &file_type)) {
return false;
}
return direct_writable_types_.find(file_type) !=
direct_writable_types_.end();
}
void SetFilesystemActiveNoLock(
bool active, IOStatus error = IOStatus::Corruption("Not active")) {
error.PermitUncheckedError();
filesystem_active_ = active;
if (!active) {
error_ = error;
}
}
void SetFilesystemActive(
bool active, IOStatus error = IOStatus::Corruption("Not active")) {
MutexLock l(&mutex_);
error.PermitUncheckedError();
SetFilesystemActiveNoLock(active, error);
}
void SetFilesystemDirectWritable(bool writable) {
MutexLock l(&mutex_);
filesystem_writable_ = writable;
}
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
// In places (e.g. GetSortedWals()) RocksDB relies on querying the file size
// or even reading the contents of files currently open for writing, and
// as in POSIX semantics, expects to see the flushed size and contents
// regardless of what has been synced. FaultInjectionTestFS historically
// did not emulate this behavior, only showing synced data from such read
// operations. (Different from FaultInjectionTestEnv--sigh.) Calling this
// function with false restores this historical behavior for testing
// stability, but use of this semantics must be phased out as it is
// inconsistent with expected FileSystem semantics. In other words, this
// functionality is DEPRECATED. Intended to be set after construction and
// unchanged (not thread safe).
void SetReadUnsyncedData(bool read_unsynced_data) {
read_unsynced_data_ = read_unsynced_data;
}
bool ReadUnsyncedData() const { return read_unsynced_data_; }
Ensure Close() before LinkFile() for WALs in Checkpoint (#12734) Summary: POSIX semantics for LinkFile (hard links) allow linking a file that is still being written two, with both the source and destination showing any subsequent writes to the source. This may not be practical semantics for some FileSystem implementations such as remote storage. They might only link the flushed or sync-ed file contents at time of LinkFile, or might even have undefined behavior if LinkFile is called on a file still open for write (not yet "sealed"). This change builds on https://github.com/facebook/rocksdb/issues/12731 to bring more hygiene to our handling of WAL files in Checkpoint. Specifically, we now Close WAL files as soon as they are either (a) inactive and fully synced, or (b) inactive and obsolete (so maybe never fully synced), rather than letting Close() happen in handling obsolete files (maybe a background thread). This should not be a performance issue as Close() should be trivial cost relative to other IO ops, but just in case: * We don't Close() while holding a mutex, to avoid blocking, and * The old behavior is available with a new kill switch option `background_close_inactive_wals`. Stacked on https://github.com/facebook/rocksdb/issues/12731 Pull Request resolved: https://github.com/facebook/rocksdb/pull/12734 Test Plan: Extended existing unit test, especially adding a hygiene check to FaultInjectionTestFS to detect LinkFile() on a file still open for writes. FaultInjectionTestFS already has relevant tracking data, and tests can opt out of the new check, as in a smoke test I have left for the old, deprecated functionality `background_close_inactive_wals=true`. Also ran lengthy blackbox_crash_test to ensure the hygiene check is OK with the crash test. (The only place I can find we use LinkFile in production is Checkpoint.) Reviewed By: cbi42 Differential Revision: D58295284 Pulled By: pdillinger fbshipit-source-id: 64d90ed8477e2366c19eaf9c4c5ad60b82cac5c6
2024-06-12 18:48:45 +00:00
// FaultInjectionTestFS normally includes a hygiene check for FileSystem
// implementations that only support LinkFile() on closed files (not open
// for write). Setting this to true bypasses the check.
void SetAllowLinkOpenFile(bool allow_link_open_file = true) {
allow_link_open_file_ = allow_link_open_file;
}
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
2021-10-11 23:22:10 +00:00
void AssertNoOpenFile() { assert(open_managed_files_.empty()); }
IOStatus GetError() { return error_; }
void SetFileSystemIOError(IOStatus io_error) {
MutexLock l(&mutex_);
io_error.PermitUncheckedError();
error_ = io_error;
}
// To simulate the data corruption before data is written in FS
void IngestDataCorruptionBeforeWrite() {
MutexLock l(&mutex_);
ingest_data_corruption_before_write_ = true;
}
void NoDataCorruptionBeforeWrite() {
MutexLock l(&mutex_);
ingest_data_corruption_before_write_ = false;
}
bool ShouldDataCorruptionBeforeWrite() {
MutexLock l(&mutex_);
return ingest_data_corruption_before_write_;
}
void SetChecksumHandoffFuncType(const ChecksumType& func_type) {
MutexLock l(&mutex_);
checksum_handoff_func_type_ = func_type;
}
const ChecksumType& GetChecksumHandoffFuncType() {
MutexLock l(&mutex_);
return checksum_handoff_func_type_;
}
void SetFailGetUniqueId(bool flag) {
MutexLock l(&mutex_);
fail_get_file_unique_id_ = flag;
}
bool ShouldFailGetUniqueId() {
MutexLock l(&mutex_);
return fail_get_file_unique_id_;
}
// Specify what the operation, so we can inject the right type of error
enum ErrorOperation : char {
kRead = 0,
kMultiReadSingleReq = 1,
kMultiRead = 2,
kOpen,
};
// Set thread-local parameters for error injection. The first argument,
// seed is the seed for the random number generator, and one_in determines
// the probability of injecting error (i.e an error is injected with
// 1/one_in probability)
void SetThreadLocalReadErrorContext(uint32_t seed, int one_in,
bool retryable) {
struct ErrorContext* ctx =
static_cast<struct ErrorContext*>(thread_local_error_->Get());
if (ctx == nullptr) {
ctx = new ErrorContext(seed);
thread_local_error_->Reset(ctx);
}
ctx->one_in = one_in;
ctx->count = 0;
ctx->retryable = retryable;
}
static void DeleteThreadLocalErrorContext(void* p) {
ErrorContext* ctx = static_cast<ErrorContext*>(p);
delete ctx;
}
// This is to set the parameters for the write error injection.
// seed is the seed for the random number generator, and one_in determines
// the probability of injecting error (i.e an error is injected with
// 1/one_in probability). For write error, we can specify the error we
// want to inject. Types decides the file types we want to inject the
// error (e.g., Wal files, SST files), which is empty by default.
void SetRandomWriteError(uint32_t seed, int one_in, IOStatus error,
bool inject_for_all_file_types,
const std::vector<FileType>& types) {
MutexLock l(&mutex_);
Random tmp_rand(seed);
error.PermitUncheckedError();
error_ = error;
write_error_rand_ = tmp_rand;
write_error_one_in_ = one_in;
inject_for_all_file_types_ = inject_for_all_file_types;
write_error_allowed_types_ = types;
}
void SetDirectWritableTypes(const std::set<FileType>& types) {
MutexLock l(&mutex_);
direct_writable_types_ = types;
}
void SetRandomMetadataWriteError(int one_in) {
MutexLock l(&mutex_);
metadata_write_error_one_in_ = one_in;
}
// If the value is not 0, it is enabled. Otherwise, it is disabled.
void SetRandomReadError(int one_in) { read_error_one_in_ = one_in; }
bool ShouldInjectRandomReadError() {
auto one_in = read_error_one_in();
return one_in > 0 && Random::GetTLSInstance()->OneIn(one_in);
}
// Inject an write error with randomlized parameter and the predefined
// error type. Only the allowed file types will inject the write error
IOStatus InjectWriteError(const std::string& file_name);
// Ingest error to metadata operations.
IOStatus InjectMetadataWriteError();
// Inject an error. For a READ operation, a status of IOError(), a
// corruption in the contents of scratch, or truncation of slice
// are the types of error with equal probability. For OPEN,
// its always an IOError.
// fault_injected returns whether a fault is injected. It is needed
// because some fault is inected with IOStatus to be OK.
IOStatus InjectThreadSpecificReadError(ErrorOperation op, Slice* slice,
bool direct_io, char* scratch,
bool need_count_increase,
bool* fault_injected);
// Get the count of how many times we injected since the previous call
int GetAndResetErrorCount() {
ErrorContext* ctx = static_cast<ErrorContext*>(thread_local_error_->Get());
int count = 0;
if (ctx != nullptr) {
count = ctx->count;
ctx->count = 0;
}
return count;
}
void EnableErrorInjection() {
ErrorContext* ctx = static_cast<ErrorContext*>(thread_local_error_->Get());
if (ctx) {
ctx->enable_error_injection = true;
}
}
void EnableWriteErrorInjection() {
MutexLock l(&mutex_);
enable_write_error_injection_ = true;
}
void EnableMetadataWriteErrorInjection() {
MutexLock l(&mutex_);
enable_metadata_write_error_injection_ = true;
}
void DisableWriteErrorInjection() {
MutexLock l(&mutex_);
enable_write_error_injection_ = false;
}
void DisableErrorInjection() {
ErrorContext* ctx = static_cast<ErrorContext*>(thread_local_error_->Get());
if (ctx) {
ctx->enable_error_injection = false;
}
}
void DisableMetadataWriteErrorInjection() {
MutexLock l(&mutex_);
enable_metadata_write_error_injection_ = false;
}
int read_error_one_in() const { return read_error_one_in_.load(); }
Fix a bug causing duplicate trailing entries in WritableFile (buffered IO) (#9236) Summary: `db_stress` is a user of `FaultInjectionTestFS`. After injecting a write error, `db_stress` probabilistically determins data drop (https://github.com/facebook/rocksdb/blob/6.27.fb/db_stress_tool/db_stress_test_base.cc#L2615:L2619). In some of our recent runs of `db_stress`, we found duplicate trailing entries corresponding to file trivial move in the MANIFEST, causing the recovery to fail, because the file move operation is not idempotent: you cannot delete a file from a given level twice. Investigation suggests that data buffering in both `WritableFileWriter` and `FaultInjectionTestFS` may be the root cause. WritableFileWriter buffers data to write in a memory buffer, `WritableFileWriter::buf_`. After each `WriteBuffered()`/`WriteBufferedWithChecksum()` succeeds, the `buf_` is cleared. If the underlying file `WritableFileWriter::writable_file_` is opened in buffered IO mode, then `FaultInjectionTestFS` buffers data written for each file until next file sync. After an injected error, user of `FaultInjectionFS` can choose to drop some or none of previously buffered data. If `db_stress` does not drop any unsynced data, then such data will still exist in the `FaultInjectionTestFS`'s buffer. Existing implementation of `WritableileWriter::WriteBuffered()` does not clear `buf_` if there is an error. This may lead to the data being buffered two copies: one in `WritableFileWriter`, and another in `FaultInjectionTestFS`. We also know that the `WritableFileWriter` of MANIFEST file will close upon an error. During `Close()`, it will flush the content in `buf_`. If no write error is injected to `FaultInjectionTestFS` this time, then we end up with two copies of the data appended to the file. To fix, we clear the `WritableFileWriter::buf_` upon failure as well. We focus this PR on files opened in non-direct mode. This PR includes a unit test to reproduce a case when write error injection to `WritableFile` can cause duplicate trailing entries. Pull Request resolved: https://github.com/facebook/rocksdb/pull/9236 Test Plan: make check Reviewed By: zhichao-cao Differential Revision: D33033984 Pulled By: riversand963 fbshipit-source-id: ebfa5a0db8cbf1ed73100528b34fcba543c5db31
2021-12-13 16:59:20 +00:00
int write_error_one_in() const { return write_error_one_in_; }
// We capture a backtrace every time a fault is injected, for debugging
// purposes. This call prints the backtrace to stderr and frees the
// saved callstack
void PrintFaultBacktrace();
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
void AddUnsyncedToRead(const std::string& fname, size_t offset, size_t n,
Slice* result, char* scratch);
private:
port::Mutex mutex_;
std::map<std::string, FSFileState> db_file_state_;
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
2021-10-11 23:22:10 +00:00
std::set<std::string> open_managed_files_;
// directory -> (file name -> file contents to recover)
// When data is recovered from unsyned parent directory, the files with
// empty file contents to recover is deleted. Those with non-empty ones
// will be recovered to content accordingly.
std::unordered_map<std::string, std::map<std::string, std::string>>
dir_to_new_files_since_last_sync_;
bool filesystem_active_; // Record flushes, syncs, writes
bool filesystem_writable_; // Bypass FaultInjectionTestFS and go directly
// to underlying FS for writable files
FaultInjectionTestFS read unsynced data by default (#12729) Summary: In places (e.g. GetSortedWals()) RocksDB relies on querying the file size or even reading the contents of files currently open for writing, and as in POSIX semantics, expects to see the flushed size and contents regardless of what has been synced. FaultInjectionTestFS historically did not emulate this behavior, only showing synced data from such read operations. (Different from FaultInjectionTestEnv--sigh.) This change makes the "proper" behavior the default behavior, at least for GetFileSize and FSSequentialFile. However, this new functionality is disabled in db_stress because of undiagnosed, unresolved issues. Also removes unused and confusing field `pos_at_last_flush_` This change is needed to support testing a relevant bug fix (in a follow-up diff). Other suggested follow-up: * Fix db_stress not to rely on the old behavior, and fix a related FIXME in db_stress_test_base.cc in LockWAL testing. * Fill in some corner cases in the FileSystem API for reading unsynced data (see new TODO items). * Consider deprecating and removing Flush() API functions from FileSystem APIs. It is not clear to me that there is a supported scenario in which they do anything but confuse API users and developers. If there is a use for them, it doesn't appear to be tested. Pull Request resolved: https://github.com/facebook/rocksdb/pull/12729 Test Plan: applies to all unit tests successfully, just updating the unit test from https://github.com/facebook/rocksdb/issues/12556 due to relying on the errant behavior. Also added a specific unit test Reviewed By: hx235 Differential Revision: D58091835 Pulled By: pdillinger fbshipit-source-id: f47a63b2b000f5875b6293a98577bff663d7fd33
2024-06-04 22:25:23 +00:00
bool read_unsynced_data_; // See SetReadUnsyncedData()
Ensure Close() before LinkFile() for WALs in Checkpoint (#12734) Summary: POSIX semantics for LinkFile (hard links) allow linking a file that is still being written two, with both the source and destination showing any subsequent writes to the source. This may not be practical semantics for some FileSystem implementations such as remote storage. They might only link the flushed or sync-ed file contents at time of LinkFile, or might even have undefined behavior if LinkFile is called on a file still open for write (not yet "sealed"). This change builds on https://github.com/facebook/rocksdb/issues/12731 to bring more hygiene to our handling of WAL files in Checkpoint. Specifically, we now Close WAL files as soon as they are either (a) inactive and fully synced, or (b) inactive and obsolete (so maybe never fully synced), rather than letting Close() happen in handling obsolete files (maybe a background thread). This should not be a performance issue as Close() should be trivial cost relative to other IO ops, but just in case: * We don't Close() while holding a mutex, to avoid blocking, and * The old behavior is available with a new kill switch option `background_close_inactive_wals`. Stacked on https://github.com/facebook/rocksdb/issues/12731 Pull Request resolved: https://github.com/facebook/rocksdb/pull/12734 Test Plan: Extended existing unit test, especially adding a hygiene check to FaultInjectionTestFS to detect LinkFile() on a file still open for writes. FaultInjectionTestFS already has relevant tracking data, and tests can opt out of the new check, as in a smoke test I have left for the old, deprecated functionality `background_close_inactive_wals=true`. Also ran lengthy blackbox_crash_test to ensure the hygiene check is OK with the crash test. (The only place I can find we use LinkFile in production is Checkpoint.) Reviewed By: cbi42 Differential Revision: D58295284 Pulled By: pdillinger fbshipit-source-id: 64d90ed8477e2366c19eaf9c4c5ad60b82cac5c6
2024-06-12 18:48:45 +00:00
bool allow_link_open_file_; // See SetAllowLinkOpenFile()
IOStatus error_;
enum ErrorType : int {
kErrorTypeStatus = 0,
kErrorTypeCorruption,
kErrorTypeTruncated,
kErrorTypeMax
};
struct ErrorContext {
Random rand;
int one_in;
int count;
bool enable_error_injection;
void* callstack;
std::string message;
int frames;
ErrorType type;
bool retryable;
explicit ErrorContext(uint32_t seed)
: rand(seed),
enable_error_injection(false),
callstack(nullptr),
frames(0),
retryable(false) {}
~ErrorContext() {
if (callstack) {
free(callstack);
}
}
};
std::unique_ptr<ThreadLocalPtr> thread_local_error_;
bool enable_write_error_injection_;
bool enable_metadata_write_error_injection_;
Random write_error_rand_;
int write_error_one_in_;
int metadata_write_error_one_in_;
std::atomic<int> read_error_one_in_;
bool inject_for_all_file_types_;
std::vector<FileType> write_error_allowed_types_;
// File types where direct writable is skipped.
std::set<FileType> direct_writable_types_;
bool ingest_data_corruption_before_write_;
ChecksumType checksum_handoff_func_type_;
bool fail_get_file_unique_id_;
// Extract number of type from file name. Return false if failing to fine
// them.
bool TryParseFileName(const std::string& file_name, uint64_t* number,
FileType* type);
};
} // namespace ROCKSDB_NAMESPACE