看代码随手记：log_put.c, __log_write()
/* * If we‘re writing the first block in a log file on a filesystem that * guarantees unwritten blocks are zero-filled, we set the size of the * file in advance. This increases sync performance on some systems, * because they don‘t need to update metadata on every sync. * * Ignore any error -- we may have run out of disk space, but that‘s no * reason to quit. */ #ifdef HAVE_FILESYSTEM_NOTZERO if (lp->w_off == 0 && !__os_fs_notzero()) { #else if (lp->w_off == 0) { #endif (void)__db_file_extend(env, dblp->lfhp, lp->log_size); if (F_ISSET(dblp, DBLOG_ZERO)) (void)__db_zero_extend(env, dblp->lfhp, 0, lp->log_size/lp->buffer_size, lp->buffer_size); }
我的理解：在flush log时使用fdatasync， 若log文件长度发生变化， 则仍需要写文件 metadata。
https://linux.die.net/man/2/fdatasync
fdatasync() is similar to fsync(), but does not flush modified metadata unless that metadata is needed in order to allow a subsequent data retrieval to be correctly handled. For example, changes to st_atime or st_mtime (respectively, time of last access and time of last modification; see stat(2)) do not require flushing because they are not necessary for a subsequent data read to be handled correctly. On the other hand, a change to the file size (st_size, as made by say ftruncate(2)), would require a metadata flush.
原文：http://www.cnblogs.com/brayden/p/6254472.html