c4e2066dbd
Summary: We can only check key on restored backup if in a stress test configuration locking the key. (Fixes mismatch seen in backup/restore with atomic flush.) TestCheckpoint used a very ugly solution to the same problem: copy-paste dozens of lines of code with some changes and removals. I removed the unnecessary implementation and made the existing one simply adaptive, like TestBackupRestore. Also made TestBackupRestore clean up dead backup/restore directories on success. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7373 Test Plan: blackbox_crash_test_with_atomic_flush for a while, blackbox_crash_test for a while, with backup and checkpoint 1 in 5k and only 1k max_keys to stress this area Reviewed By: ajkr Differential Revision: D23629057 Pulled By: pdillinger fbshipit-source-id: d7fe7e2be75aaf3cf974be9540a7c5c5de8b371b |
||
---|---|---|
.. | ||
batched_ops_stress.cc | ||
cf_consistency_stress.cc | ||
CMakeLists.txt | ||
db_stress_common.cc | ||
db_stress_common.h | ||
db_stress_compaction_filter.h | ||
db_stress_driver.cc | ||
db_stress_driver.h | ||
db_stress_env_wrapper.h | ||
db_stress_gflags.cc | ||
db_stress_listener.h | ||
db_stress_shared_state.cc | ||
db_stress_shared_state.h | ||
db_stress_stat.h | ||
db_stress_table_properties_collector.h | ||
db_stress_test_base.cc | ||
db_stress_test_base.h | ||
db_stress_tool.cc | ||
db_stress.cc | ||
no_batched_ops_stress.cc |