#!/usr/bin/ksh
#
# (c) 2009, lists@nerdbynature.de
#
# Yet another filesystem benchmark script
#
# - bonnie++
# - dbench
# - iozone3
# - tiobench
# - some generic operations, like tar/touch/rm
#
# TODO:
# - different mount options for each filesystems
# - different benchmark options
#
# v0.1 - initial version
# v0.2 - disabled 2 filesystems
#        ufs - http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=526586
#        nilfs2 - filesystem fills up until ENOSPC
# v0.3 - run tiobench with only 1 thread, otherwise we get:
#        Illegal division by zero at /usr/bin/tiobench line 163
# v0.4 - rewrite for ksh: this way we have fancy features to replace seq(1)
#        but still don't have to use a full blown bash.
#         
# 
CONF="/usr/local/etc/fs-bench.conf"

log() {
echo "`date +'%F %H:%M:%S'`: $1"
[ -n "$2" ] && exit "$2"
}

# sanity checks
if [ ! -b "$1" -o ! -d "$2" -o ! -f $CONF ]; then
	log "Usage: `basename $0` [dev] [mpt]"
	log "Make sure $CONF exists!" 1
else
	DEV="$1"
	MPT="`echo $2 | sed 's/\/$//'`"
	. "$CONF"
fi

# overwrite results dir?
if [ -d "$LOG" ]; then
	printf "Directory $LOG already exists, overwrite? (y/n) " && read c
	if [ $c = "y" ]; then
		$DEBUG rm -rf "$LOG"
		$DEBUG mkdir -p "$LOG"/raw
	else
		log "Aborted." 1
	fi
else
	$DEBUG mkdir "$LOG"
fi

########################################################
mkfs_btrfs() {
$DEBUG mkfs.btrfs $DEV 1>/dev/null
$DEBUG mount -t btrfs -o noatime $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_ext2() {
$DEBUG mkfs.ext2 -q $DEV
$DEBUG mount -t ext2 -o noatime,user_xattr $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_ext3() {
$DEBUG mkfs.ext3 -q $DEV
$DEBUG mount -t ext3 -o noatime,user_xattr $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_ext4() {
$DEBUG mkfs.ext4 -q $DEV
$DEBUG mount -t ext4 -o noatime,user_xattr $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_jfs() {
$DEBUG mkfs.jfs -q $DEV 1>/dev/null
$DEBUG mount -t jfs -o noatime $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_nilfs2() {
$DEBUG mkfs.nilfs2 $DEV 2>/dev/null
$DEBUG mount -t nilfs2 -o noatime $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_reiserfs() {
$DEBUG mkfs.reiserfs -q $DEV > /dev/null 2>&1
$DEBUG mount -t reiserfs -o noatime,user_xattr $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_ufs() {
$DEBUG mkfs.ufs -O2 -U $DEV > /dev/null 2>&1
$DEBUG mount -t ufs -o noatime $DEV $MPT > /dev/null 2>&1
ERR=$?
}

mkfs_xfs() {
$DEBUG mkfs.xfs -q -f $DEV
$DEBUG mount -t xfs -o noatime $DEV $MPT > /dev/null 2>&1
ERR=$?
}

########################################################
run_bonnie() {
eval conf_bonnie
log "Running $b on $fs..."
$DEBUG bonnie++ -u root -r $RAM -s $SIZE -m "$fs" -x 1 -d "$MPT" 1>$LOG/raw/bonnie-$fs.csv 2>$LOG/raw/bonnie-$fs.err
$DEBUG cat $LOG/raw/bonnie-*.csv | bon_csv2html > $LOG/bonnie.html
}

########################################################
run_stress() {
#
# Based on http://oss.oracle.com/~mason/stress.sh
#      Copyright (C) 1999 Bibliotech Ltd., 631-633 Fulham Rd., London SW6 5UQ.
#      $Id: stress.sh,v 1.2 1999/02/10 10:58:04 rich Exp $
#
eval conf_stress
log "Running $b on $fs (size: `du -sk "$CONTENT" | awk '{print $1 / 1024 " MB"}'`)..."
$DEBUG mkdir $MPT/stress || log "cannot create $MPT/stress" 1
$DEBUG    cd $MPT/stress || log "cannot cd into $MPT/stress" 1

# computing MD5 sums over content directory
find $CONTENT -type f -exec md5sum '{}' + | sort > $MPT/content.sums

# starting stress test processes
p=1
while [ $p -le $CONCURRENT ]; do
	(
	# wait for all processes to start up.
	if [ "$STAGGER" = "yes" ]; then
		sleep `expr 2 \* $p`
	else
		sleep 1
	fi

	r=1
	while [ $r -le $RUNS ]; do
		log "Running stresstest in $MPT/stress/$p (r: $r)..."
		# Remove old directories.
		$DEBUG rm -rf $MPT/stress/$p

		# copy content
		$DEBUG mkdir $MPT/stress/$p || log "cannot create $MPT/stress/$p"
		$DEBUG cp -dRx $CONTENT/* $MPT/stress/$p || log "cannot copy $CONTENT to $MPT/stress/$p"

		# compare the content and the copy.
		$DEBUG cd $MPT/stress/$p
		$DEBUG find . -type f -exec md5sum '{}' + | sort > $MPT/stress.$p
		$DEBUG diff -q $MPT/content.sums $MPT/stress.$p
		if [ $? != 0 ]; then
			log "corruption found in $MPT/stress/$p (r: $r)"
			continue
		fi
		$DEBUG cd $MPT/stress
		$DEBUG rm -f $MPT/stress.$p
	r=`expr $r + 1`
	done
	) &
p=`expr $p + 1`
done
}

########################################################
run_dbench() {
eval conf_dbench
log "Running $b on $fs..."
$DEBUG dbench -x -t $TIME -D $MPT $NPROC > $LOG/raw/dbench-$fs.log
echo "$fs:	`egrep '^Throughput' $LOG/raw/dbench-$fs.log`" >> $LOG/dbench.txt
}

run_iozone() {
eval conf_iozone
log "Running $b on $fs..."
$DEBUG cd $MPT || log "cannot cd into $MPT" 1
$DEBUG iozone -a -c -S $CACHESIZE -s $FILESIZE > $LOG/raw/iozone-$fs.log
}

run_tiobench() {
eval conf_tiobench
log "Running $b on $fs..."
$DEBUG tiobench --identifier fs_"$fs" --size $SIZE --numruns $NUMRUNS --dir $MPT \
	--block 4096 --block 8192 --threads 1 \
	1>$LOG/raw/tiobench-$fs.log 2>$LOG/raw/tiobench-$fs.err

# results are hard to summarize
echo "                               File  Blk   Num                   Avg      Maximum      Lat%     Lat%    CPU" >  $LOG/tiobench.txt
echo "                               Size  Size  Thr   Rate  (CPU%)  Latency    Latency      >2s      >10s    Eff" >> $LOG/tiobench.txt
for t in "Sequential Reads" "Sequential Writes" "Random Reads" "Random Writes"; do
	echo $t
	# adjust -An for more/less than 2 different blocksizes!
	grep -h -A5 "$t" $LOG/raw/tiobench-*.log | egrep '^fs_'
	echo
done >> $LOG/tiobench.txt
}

run_generic() {
eval conf_generic
CONTENT_SIZE=`du -sk "$CONTENT" | awk '{print $1}'`
ERR=""
log "Running $b on $fs..."
# - copy content to fs
# - tar up local content
# - copy content within same fs
# - create NUMFILES in NUMDIRS

GEN_BEGIN=`date +%s`
$DEBUG tar -cf - $CONTENT 2>/dev/null | tar -C $MPT -xf - 2>/dev/null
$DEBUG diff -q $CONTENT $MPT/$CONTENT > /dev/null 2>&1
[ $? = 0 ] || ERR="- FAILED"
$DEBUG sync
GEN_END=`date +%s`
GEN_DUR=`echo "scale=2; $GEN_END - $GEN_BEGIN" | bc -l`
  SPEED=`echo "scale=2; ( $CONTENT_SIZE / $GEN_DUR ) / 1024" | bc -l`
log "$fs: 1-tar content to $MPT and running diff took $GEN_DUR seconds. ($SPEED MB/s) $ERR"  > $LOG/raw/generic-$fs.log

GEN_BEGIN=`date +%s`
cd $MPT
$DEBUG tar -cf - . 2>/dev/null | dd of=/dev/null 2>/dev/null
[ $? = 0 ] || ERR="- FAILED"
$DEBUG sync
GEN_END=`date +%s`
GEN_DUR=`echo "scale=2; $GEN_END - $GEN_BEGIN" | bc -l`
  SPEED=`echo "scale=2; ( $CONTENT_SIZE / $GEN_DUR ) / 1024" | bc -l`
log "$fs: 2-tar content on $MPT took $GEN_DUR seconds. ($SPEED MB/s) $ERR" >> $LOG/raw/generic-$fs.log


GEN_BEGIN=`date +%s`
cd $MPT
$DEBUG mkdir TEST.$$
$DEBUG cp -fpR * TEST.$$ 2>/dev/null
[ $? = 0 ] || ERR="- FAILED"
$DEBUG sync
GEN_END=`date +%s`
GEN_DUR=`echo "scale=2; $GEN_END - $GEN_BEGIN" | bc -l`
  SPEED=`echo "scale=2; ( $CONTENT_SIZE / $GEN_DUR ) / 1024" | bc -l`
log "$fs: 3-cp content took $GEN_DUR seconds. ($SPEED MB/s) $ERR" >> $LOG/raw/generic-$fs.log

# create many files
GEN_BEGIN=`date +%s`
$DEBUG mkdir -p $MPT/manyfiles
for d in `seq 1 $NUMDIRS`; do
	$DEBUG mkdir -p $MPT/manyfiles/dir."$d" 2>>$LOG/raw/generic-$fs.err
	$DEBUG cd $MPT/manyfiles/dir."$d"       2>>$LOG/raw/generic-$fs.err
	seq 1 $NUMFILES | xargs touch           2>>$LOG/raw/generic-$fs.err
done
sync
GEN_END=`date +%s`
GEN_DUR=`echo "scale=2; $GEN_END - $GEN_BEGIN" | bc -l`
 NUMDIRS_CREATED=`cd $MPT/manyfiles; find . -type d | egrep -v '^\.$' | wc -l`
NUMFILES_CREATED=`cd $MPT/manyfiles; find . -type f | wc -l`
[ $NUMDIRS_CREATED = $NUMDIRS -a $NUMFILES_CREATED = $NUMFILES ] || ERR="- FAILED"
log "$fs: 4-creating $NUMFILES files in each of the $NUMDIRS directories took $GEN_DUR seconds. $ERR" >> $LOG/raw/generic-$fs.log

# create many dirs (we just replace NUMDIRS with NUMFILES and vice versa)
GEN_BEGIN=`date +%s`
$DEBUG mkdir -p $MPT/manydirs
for d in `seq 1 $NUMFILES`; do
	$DEBUG mkdir -p $MPT/manydirs/dir."$d" 2>>$LOG/raw/generic-$fs.err
	$DEBUG cd $MPT/manydirs/dir."$d"       2>>$LOG/raw/generic-$fs.err
	seq 1 $NUMDIRS | xargs touch           2>>$LOG/raw/generic-$fs.err
done
sync
GEN_END=`date +%s`
GEN_DUR=`echo "scale=2; $GEN_END - $GEN_BEGIN" | bc -l`
 NUMDIRS_CREATED=`cd $MPT/manydirs; find . -type d | egrep -v '^\.$' | wc -l`
NUMFILES_CREATED=`cd $MPT/manydirs; find . -type f | wc -l`
[ $NUMDIRS_CREATED = $NUMFILES -a $NUMFILES_CREATED = $NUMDIRS ] || ERR="- FAILED"
log "$fs: 5-creating $NUMDIRS files in each of the $NUMFILES directories took $GEN_DUR seconds. $ERR" >> $LOG/raw/generic-$fs.log

# results across all tests
echo "Content size is `echo "scale=2; $CONTENT_SIZE / 1024" | bc -l` MB" > $LOG/generic.txt
for t in `seq 1 5`; do
	grep -h -- "$t"- $LOG/raw/generic-*.log | sed 's/^.*[0-9]: //;s/[0-9]-//'
	echo
done >> $LOG/generic.txt
}

########################################################
mount | grep -q "$MPT" && log "$MPT is already mounted!" 1

BEGIN=`date +%s`
for fs in $FILESYSTEMS; do
	echo "========================================================"
	FS_BEG=`date +%s`
	# mkfs, mount, umount for every benchmark
	for b in $BENCHMARKS; do
		mkfs_"$fs"
		if [ ! $ERR = 0 ]; then
			log "mkfs failed ($fs, $b)"
			continue
		fi
		BM_BEG=`date +%s`
		run_"$b"
		# 2>&1 | tee "$LOG"/"$b"-"$fs".log
		$DEBUG cd /
		$DEBUG sync
		$DEBUG sleep 5
		$DEBUG sync
		$DEBUG umount $MPT || log "Unmounting $fs failed!" 1
		BM_END=`date +%s`
		log "Running $b on $fs took `echo "scale=2; ( $BM_END - $BM_BEG ) / 60" | bc -l` minutes."
	done
	FS_END=`date +%s`
	echo
	log "Running all benchmarks on $fs took `echo "scale=2; ( $FS_END - $FS_BEG ) / 60" | bc -l` minutes."
	echo
done 2>&1 | tee $LOG/fs-bench.log

END=`date +%s`
log "Finished after `echo "scale=2; ( $END - $BEGIN ) / 60" | bc -l` minutes."