The high-performance distributed file system FastCFS 5.4 has been released, mainly for performance improvement and bug fix.
Performance improvement
libdiskallocator has been optimized for faststore performance: when recycling trunk for data migration, adjacent slices are merged and the number of data reads is reduced; When a trunk space is allocated and data is written, merge adjacent slices to reduce the number of slices.
Small improvements
[libfastcommon] impl. function shorten_path for /./ and /../
[libserverframe] sf_binlog_writer_rotate_file can skip empty file
[fstore] storage.conf and dbstore.conf changed for the newest libdiskallocator
bug fixed
[libserverframe] call flush_writer_files before rotate file
[libdiskallocator] da_init_start split to init and start 2 functions
[libfdirstorage] set last segment inode range correctly
[libfdirstorage] keep for update correctly
[fdir] call fdir_namespace_inc_alloc_bytes correctly