Copied from
ubuntu lucid in
Primary Archive for Ubuntu
Changelog
bonnie++ (1.96) unstable; urgency=low
* Fix a minor bug in setting the GID.
* Make it build on Solaris by checking for O_DIRECT being defined.
* The option -n9765 or greater caused the small file test to abort,
to increase this I changed the number format to hexadecimal and made it
10 digits. Now any option smaller than -n1073741824 will be OK.
Thanks Nick Dokos of HP.
* Updated bon_csv2html, made the machine name span two rows, made the
concurrency column go away when there's no data, correctly calculated the
colors of the speed columns (was off by one column before), and calculate
the CPU colors.
bonnie++ (1.95) unstable; urgency=low
* Removed code for building on OS/2, I think that no-one has been interested
in this for a while.
* Removed code for building on NT, I don't think that anyone was ever very
interested in that and it probably didn't work.
* Patch from Justin Clift to make getc_putc compile on RHEL 5.2.
* Added the -D option to bonnie++ to use direct IO (O_DIRECT) for bulk IO.
Thanks to Dave Murch from Violin Memory - http://violin-memory.com/
* Fixed some more warnings with recent versions of GCC.
bonnie++ (1.94) unstable; urgency=low
* Fixed typos in man page.
Closes: #133714, #449596
* Added the URL to the man pages.
* Documented why UID==0 is bad.
Closes: #182023
* Check return value of chdir()
* Removed the zcav -s option which did the same thing as -n.
* Made zcav report in GiB and MiB/s, and made the default 256MiB.
* Improved bonnie++.8 documentation of the -y and -p options.
* Made zcav support skipping blocks for faster completion.
* Fixed the autoconf checks for 64bit seek.
bonnie++ (1.93d) unstable; urgency=low
* Documented the default chunk size in bonnie++.8.
* Added support for testing write speed to zcav.
* Made default block size for zcav 200M.
bonnie++ (1.93c) unstable; urgency=low
* Made it compile with GCC 3.2.
bonnie++ (1.93b) unstable; urgency=low
* Fixed an error in calculating seeks, it used to add an extra 5 seconds to
the test time before calculation and therefore under-reported the results.
* Fixed the signal handling error on Irix.
* Fixed <algo> check to work with GCC 3.2.
* Fixed a bug where it would SEGV if you specified more than 999 directories
(now it handles 99,999 and displays an error message if you ask for more).
* Fixed a bug in error handling during a stat test with multiple directories.
* Fixed the mandir for RPM package building.
bonnie++ (1.93a) unstable; urgency=low
* Made the signal handler extern "C".
* Updated the RPM packaging.
* Improved the NT portability code.
bonnie++ (1.93) unstable; urgency=low
* Updated to latest debian standards version and follow the latest Debian
policy on stripping and debug generation.
* Removed optind=0 because it caused problems on BSD.
* Added new test program getc_putc for per char IO.
bonnie++ (1.92b) unstable; urgency=low
* Fixed a buffer overflow in the random number management code.
* Fixed a bug where bonnie++ would SEGV if ^C was pressed at the start of the
file creation tests.
* Clarified the -r option to bonnie++ in the man page.
* Finished adding the support for specifying the seed. Fixed the bon_csv2*
programs and their documentation.
bonnie++ (1.92a) unstable; urgency=low
* Made the -d option to bonnie++ work with relative paths. Now it does
chdir() to the specifed directory and proceeds as usual.
* Fixed a tricky bug related to thread management which caused SEGV. For
some time I had known of the existance of the bug but I couldn't work
out how to reproduce it.
* Added a feature of specifying the random number seed for repeatable tests.
Also made a separate class to manage all random numbers as part of this.
* Made bonnie++ have the facility to read from a file containing random
numbers and created a program to generate them.
* Added a "make dep" rule and include it's output in the archive.
* Now I don't use stupid macro names like _close which break on BSD!
bonnie++ (1.92) unstable; urgency=low
* Made the per-byte code use read() and write() instead of putc() and getc().
Using the stream functions gives too much variance and isn't as interesting
nowadays. I will soon add a separate benchmark for streams vs write() for
byte at a time.
* Made it use a single file >2G on systems that have LFS support. On systems
that lack it there will be no more support of more than 2G of data.
I believe that now any machine with 1G of RAM should have LFS support!
Note that the ./configure test for this takes ages on ReiserFS.
* Made it remove the ./configure cache files before Debian package build.
This fixes the problems with incorrect data being cached for ports.
* Fixed the memory checks for file creation tests, thanks to
Andreas Dilger <email address hidden> .
* Fixed a minor field alignment bug in bon_csv2html.
* Made bon_csv2html not display columns that contain no data.
* Fixed a bug where bon_csv2html could crash on bad data.
* Fixed a bug where bonnie++ would generate bad CSV data if run with "-n0".
bonnie++ (1.91c) unstable; urgency=low
* Now compiles with GCC 3.0.
* Stopped it giving a SIGFPE on Alpha and tidied up the relevant code in the
results printing.
* Fixed the CSV output.
* Fixed a minor warning.
bonnie++ (1.91b) unstable; urgency=low
* Added an fflush(NULL) to the Sync code to make it display the prompt on
buffered devices.
* Added autoconf check for snprintf().
* Fixed some minor warnings.
bonnie++ (1.91a) unstable; urgency=low
* Fixed the -p/-y options.
* Made -n take "k" suffix for kilobytes for the size and the "m" suffix
for creating millions of files. Also allow specifying the chunk size
for small file tests.
* Changed readme.html to make it more up to date.
* Made it conflict/provide/replace the bonnie package. Bonnie++ does
everything that bonnie does and fixes some bugs including doing srand()
after fork(), and having sensible defaults and detection of RAM installed
to avoid meaningless test results.
There is no need to have bonnie and bonnie++ installed.
* Use @exec_prefix@ in autoconf for better standards compliance.
* Made it possible to specify chunk_size for IO and file tests separately.
Added a new field in position 21 in the csv file to record this. So I
changed the compatibility number for the bon_csv2* programs.
* Removed the "wait" code in zcav which caused all the broken-ness. It
probably didn't gain us much anyway.
bonnie++ (1.91) unstable; urgency=low
* Fixed the bon_csv2html bug that made the columns not line up if different
rows had different numbers of blank columns.
* Fixed a bug introduced in 1.90f which resulted in Random Seeks always
being reported as 0.
* Made quiet mode not tell you which UID it uses.
bonnie++ (1.90g) unstable; urgency=low
* Added latency parsing support to bon_csv2html.
* Fixed a number of bugs in bon_csv2html, now does colors correctly.
bonnie++ (1.90f) unstable; urgency=low
* Now use mode "b" for fopen() for NT and OS/2.
* Made the latency numbers print out correctly.
* Added latency parsing support to bon_csv2txt.
bonnie++ (1.90e) unstable; urgency=low
* Now should compile properly on NT and OS/2 (NB ZCAV still won't fully work).
bonnie++ (1.90d) unstable; urgency=low
* Updated standards version to 3.2.1.
* Done some serious work on porting to OS/2 and NT. This may break compile
on some versions of UNIX. If so I'll fix it as soon as it gets reported.
bonnie++ (1.90c) unstable; urgency=low
* Rewrote bon_csv2html in C++ and made it assign colors to the fields to show
how fast or slow the values are (red for slow, green for fast).
bonnie++ (1.90b) unstable; urgency=low
* Added first stage of support for synchronising bonnie++ instances over a
network.
This required changing the way the -y option operates.
Also created a new Sync class to do this generically.
* Added code to check the latency of operations in micro-seconds (us).
bonnie++ (1.90a) unstable; urgency=low
* This is the start of the new 2.00 series of Bonnie++. The aim is to do
serious performance testing of RAID arrays. All programs will be multi-
threaded.
* Added the multi-threaded zcav that I had removed from the 1.00 series and
make Bonnie++ use threads instead of fork(). Next step is to add extra
concurency through threads.
* Changed the format of the csv files. Added a format version number and a
place for the version of Bonnie++ that created the data. Made the
bon_csv2txt and bon_csv2html programs understand the new format.
-- Artur Rona <email address hidden> Mon, 21 Dec 2009 18:13:38 +0000