Commit 0f24c9fc authored by heikki@donna.mysql.fi's avatar heikki@donna.mysql.fi
Browse files

os0file.c Use O_SYNC if possible, second choice fdatasync, and last choice fsync 

configure.in	Check if fdatasync exists
manual.texi	Updated defragmenting doc
manual.texi	Corrected fillfactor contract threashold for a page to 1/2
parent 265164db

Docs/manual.texi

+11 −4
Original line number Diff line number Diff line
@@ -25283,7 +25283,7 @@ of the index records. 
If index records are inserted in a sequential (ascending or descending)

order, the resulting index pages will be about 15/16 full.

If records are inserted in a random order, then the pages will be

1/2 - 15/16 full. If the fillfactor of an index page drops below 1/4,

1/2 - 15/16 full. If the fillfactor of an index page drops below 1/2,

InnoDB will try to contract the index tree to free the page.















@subsubsection Insert buffering










@@ -25440,13 +25440,20 @@ consistent read.



























If there are random insertions or deletions















in the indexes of a table, the indexes













may become fragmented. By this we mean that the physical ordering













may become fragmented. By fragmentation we mean that the physical ordering















of the index pages on the disk is not close to the alphabetical ordering













of the records on the pages. It can speed up index scans if you













of the records on the pages, or that there are many unused pages in the













64-page blocks which were allocated to the index.

























It can speed up index scans if you















periodically use @code{mysqldump} to dump the table to















a text file, drop the table, and reload it from the dump.













Another way to do the defragmenting is to @code{ALTER} the table type to













@code{MyISAM} and back to @code{InnoDB} again.













Note that a @code{MyISAM} table must fit in a single file













on your operating system.



























Note that if the insertions to and index are always ascending and













If the insertions to and index are always ascending and















records are deleted only from the end, then the the file space management















algorithm of InnoDB guarantees that fragmentation in the index will















not occur.

























innobase/configure.in



+1
−0






























Original line number
Diff line number
Diff line








@@ -37,6 +37,7 @@ AC_PROG_INSTALL













AC_CHECK_HEADERS(aio.h sched.h)















AC_CHECK_SIZEOF(int, 4)















AC_CHECK_FUNCS(sched_yield)













AC_CHECK_FUNCS(fdatasync)















#AC_C_INLINE  Already checked in MySQL















AC_C_BIGENDIAN

















































innobase/os/os0file.c



+18
−1






























Original line number
Diff line number
Diff line








@@ -299,6 +299,13 @@ os_file_create(





























	UT_NOT_USED(purpose);





























	/* On Linux opening a file in the O_SYNC mode seems to be much













        more efficient than calling an explicit fsync or fdatasync after













        each write */



























#ifdef O_SYNC













	create_flag = create_flag | O_SYNC;













#endif















	if (create_mode == OS_FILE_CREATE) {















	        file = open(name, create_flag, S_IRUSR | S_IWUSR | S_IRGRP















			                     | S_IWGRP | S_IROTH | S_IWOTH);










@@ -492,8 +499,18 @@ os_file_flush(













#else















	int	ret;















	













	ret = fsync(file);













#ifdef O_SYNC













	/* We open all files with the O_SYNC option, which means there













        should be no need for fsync or fdatasync. In practice such a need













	may be because on a Linux Xeon computer "donna" the OS seemed to be













	fooled to believe that 500 disk writes/second are possible. */





























	ret = 0;













#elif defined(HAVE_FDATASYNC)













	ret = fdatasync(file);













#else













	ret = fsync(file);













#endif















	if (ret == 0) {















		return(TRUE);















	}