问题描述
我关注上一篇文章combine parallel and gfind
我想构建gupdatedb数据库,该数据库包含来自主根/
的所有数据库,但下面列出的PRUNEPATHS
除外。我正在MacOS 10.15 Catalina上工作。
因此,我尝试在MacOS 10.15上修改gupdatedb脚本,以从parallel
命令中受益(请注意# : A2
部分):
# : A2
cat | parallel -j32 $find {} $SEARCHPATHS $FINDOPTIONS \
\( $prunefs_exp -type d -regex "$PRUNEREGEX" \) \
-prune -o $print_option * :::
如果我不使用cat |
,则会收到以下警告消息:
parallel: Warning: Input is read from the terminal. You are either an expert
parallel: Warning: (in which case: YOU ARE AWESOME!) or maybe you forgot
parallel: Warning: ::: or :::: or -a or to pipe data into parallel. If so
parallel: Warning: consider going through the tutorial: man parallel_tutorial
parallel: Warning: Press CTRL-D to exit.
并且该过程似乎挂起。
不幸的是,$find = gfind
的多个线程似乎不能同时运行:
我已经启动了这样的脚本:sudo time gupdatedb
并在以下结果以下:ps aux | grep find
:
root 84865 0.0 0.0 4459044 15828 s002 S+ 1:43PM 0:00.10 perl /usr/local/bin/parallel -j32 /usr/local/Cellar/findutils/4.7.0/bin/gfind {} / ( -fstype 9P -o -fstype NFS -o -fstype afs -o -fstype autofs -o -fstype cifs -o -fstype coda -o -fstype devfs -o -fstype devpts -o -fstype ftpfs -o -fstype iso9660 -o -fstype mfs -o -fstype ncpfs -o -fstype nfs -o -fstype nfs4 -o -fstype proc -o -fstype shfs -o -fstype smbfs -o -fstype sysfs -o -type d -regex \(^/afs$\)\|\(^/amd$\)\|\(^/proc$\)\|\(^/sfs$\)\|\(^/tmp$\)\|\(^/usr/tmp$\)\|\(^/var/tmp$\)\|\(^/Volumes$\) ) -prune -o -print0 Applications Library System Users Volumes bin cores dev etc home opt private sbin tmp usr var :::
root 84863 0.0 0.0 4268280 796 s002 S+ 1:43PM 0:00.00 /usr/local/Cellar/findutils/4.7.0/libexec/gfrcode -0
root 84861 0.0 0.0 4282172 708 s002 S+ 1:43PM 0:00.00 /bin/sh /usr/local/Cellar/findutils/4.7.0/libexec/bin/gupdatedb
root 84853 0.0 0.0 4273980 1164 s002 S+ 1:43PM 0:00.01 /bin/sh /usr/local/Cellar/findutils/4.7.0/libexec/bin/gupdatedb
root 84850 0.0 0.0 5396228 10288 s008 S+ 1:43PM 0:00.27 vim /usr/local/Cellar/findutils/4.7.0/libexec/bin/gupdatedb
root 84849 0.0 0.0 4788896 6740 s008 S+ 1:43PM 0:00.03 sudo vim /usr/local/Cellar/findutils/4.7.0/libexec/bin/gupdatedb
最后,可能无法构建数据库,我正在检查/usr/local/var/locate/locatedb.n
和/usr/local/var/locate/locatedb
的大小,但没有任何变化。
与parallel一起使用的语法有什么问题? (特别是,我不知道如何处理命令的... ::: options
部分)
PS:我已设置为gupdatedb
:
# Directories to not put in the database,which would otherwise be.
: ${PRUNEPATHS="
/afs
/amd
/proc
/sfs
/tmp
/usr/tmp
/var/tmp
/Volumes
"}
和
# You can set these in the environment,or use command-line options,# to override their defaults:
# Any global options for find?
: ${FINDOPTIONS=}
# What shell shoud we use? We should use a POSIX-ish sh.
: ${SHELL="/bin/sh"}
# Non-network directories to put in the database.
: ${SEARCHPATHS="/"}
更新1
更准确地说,这是我要求与parallel/find
对夫妇进行潜在优化(并行化)的帖子:
example of a potential parallelization with coupled parallel/find
除了脚本gupdatedb
,我想做同样的优化。
更新2
我遵循了:
的建议关于我的问题,进入gupdatedb
的defaut命令是:
$find $SEARCHPATHS $FINDOPTIONS \
\( $prunefs_exp \
-type d -regex "$PRUNEREGEX" \) -prune -o $print_option
所以,我刚刚做了如下修改:
parallel -j32 $find {} $SEARCHPATHS $FINDOPTIONS \
\( $prunefs_exp \
-type d -regex "$PRUNEREGEX" \) -prune -o $print_option ::: /
,我收到以下错误消息:
/bin/sh: -c: line 0: Syntax error near unexpected token `('
/bin/sh: -c: line 0: `/usr/local/Cellar/findutils/4.7.0/bin/gfind / / ( -fstype 9P -o -fstype NFS -o -fstype afs -o -fstype autofs -o -fstype cifs -o -fstype coda -o -fstype devfs -o -fstype devpts -o -fstype ftpfs -o -fstype iso9660 -o -fstype mfs -o -fstype ncpfs -o -fstype nfs -o -fstype nfs4 -o -fstype proc -o -fstype shfs -o -fstype smbfs -o -fstype sysfs -o -type d -regex \(^/private/tmp$\)\|\(^/private/var/folders$\)\|\(^/private/var/tmp$\)\|\(^*/Backups.backupdb$\)\|\(^/System$\)\|\(^/Volumes$\) ) -prune -o -print0'
这可能是什么问题?
更新3:
这里是脚本gupdatedb
,您可以从第300行看到我的不同尝试:
#! /bin/sh
# updatedb -- build a locate pathname database
# copyright (C) 1994-2019 Free Software Foundation,Inc.
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation,either version 3 of the License,or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or fitness FOR A PARTIculaR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not,see <https://www.gnu.org/licenses/>.
# csh original by James Woods; sh conversion by David MacKenzie.
#exec 2> /tmp/updatedb-trace.txt
#set -x
version='
updatedb (GNU findutils) 4.7.0
copyright (C) 1994-2019 Free Software Foundation,Inc.
License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY,to the extent permitted by law.
Written by Eric B. Decker,James Youngman,and Kevin Dalley.
'
# File path names are not actually text,anyway (since there is no
# mechanism to enforce any constraint that the basename of a
# subdirectory has the same character encoding as the basename of its
# parent). The practical effect is that,depending on the way a
# particular system is configured and the content of its filesystem,# passing all the file names in the system through "sort" may generate
# character encoding errors in text-based tools like "sort". To avoid
# this,we set LC_ALL=C. This will,presumably,not work perfectly on
# systems where LC_ALL is not the way to do locale configuration or
# some other seting can override this.
LC_ALL=C
export LC_ALL
# We can't use substitution on PACKAGE_URL below because it
# (correctly) points to https://www.gnu.org/software/findutils/ instead
# of the bug reporting page.
usage="\
Usage: $0 [--findoptions='-option1 -option2...']
[--localpaths='dir1 dir2...'] [--netpaths='dir1 dir2...']
[--prunepaths='dir1 dir2...'] [--prunefs='fs1 fs2...']
[--output=dbfile] [--netuser=user] [--localuser=user]
[--dbformat] [--version] [--help]
Please see also the documentation at http://www.gnu.org/software/findutils/.
Report (and track progress on fixing) bugs in the updatedb
program via the GNU findutils bug-reporting page at
https://savannah.gnu.org/bugs/?group=findutils or,if
you have no web access,by sending email to <bug-findutils@gnu.org>.
"
changeto=/
for arg
do
# If we are unable to fork,the back-tick operator will
# fail (and the shell will emit an error message). When
# this happens,we exit with error value 71 (EX_OSERR).
# Alternative candidate - 75,EX_TEMPFAIL.
opt=`echo $arg|sed 's/^\([^=]*\).*/\1/'` || exit 71
val=`echo $arg|sed 's/^[^=]*=\(.*\)/\1/'` || exit 71
case "$opt" in
--findoptions) FINDOPTIONS="$val" ;;
--localpaths) SEARCHPATHS="$val" ;;
--netpaths) NETPATHS="$val" ;;
--prunepaths) PRUNEPATHS="$val" ;;
--prunefs) PRUNEFS="$val" ;;
--output) LOCATE_DB="$val" ;;
--netuser) NETUSER="$val" ;;
--localuser) LOCALUSER="$val" ;;
--changecwd) changeto="$val" ;;
--dbformat) dbformat="$val" ;;
--version) fail=0; echo "$version" || fail=1; exit $fail ;;
--help) fail=0; echo "$usage" || fail=1; exit $fail ;;
*) echo "updatedb: invalid option $opt
Try '$0 --help' for more information." >&2
exit 1 ;;
esac
done
frcode_options=""
case "$dbformat" in
"")
# Default,use LOCATE02
;;
LOCATE02)
;;
slocate)
frcode_options="$frcode_options -S 1"
;;
*)
# The "old" database format is no longer supported.
echo "Unsupported locate database format ${dbformat}: Supported formats are:" >&2
echo "LOCATE02,slocate" >&2
exit 1
esac
if true
then
sort="/usr/bin/sort -z"
print_option="-print0"
frcode_options="$frcode_options -0"
else
sort="/usr/bin/sort"
print_option="-print"
fi
getuid() {
# format of "id" output is ...
# uid=1(daemon) gid=1(other)
# for `id's that don't understand -u
id | cut -d'(' -f 1 | cut -d'=' -f2
}
# figure out if su supports the -s option
select_shell() {
if su "$1" -s $SHELL -c false < /dev/null ; then
# No.
echo ""
else
if su "$1" -s $SHELL -c true < /dev/null ; then
# Yes.
echo "-s $SHELL"
else
# su is unconditionally failing. We won't be able to
# figure out what is wrong,so be conservative.
echo ""
fi
fi
}
# You can set these in the environment,# to override their defaults:
# Any global options for find?
: ${FINDOPTIONS="-mindepth 1 -maxdepth 1"}
#: ${FINDOPTIONS=""}
# What shell shoud we use? We should use a POSIX-ish sh.
: ${SHELL="/bin/sh"}
# Non-network directories to put in the database.
: ${SEARCHPATHS="/"}
# Network (NFS,AFS,RFS,etc.) directories to put in the database.
: ${NETPATHS=}
# Directories to not put in the database,which would otherwise be.
: ${PRUNEPATHS="
/afs
/amd
/proc
/sfs
/tmp
/usr/tmp
/var/tmp
"}
# Trailing slashes result in regex items that are never matched,which
# is not what the user will expect. Therefore we Now reject such
# constructs.
for p in $PRUNEPATHS; do
case "$p" in
/*/) echo "$0: $p: pruned paths should not contain trailing slashes" >&2
exit 1
esac
done
# The same,in the form of a regex that find can use.
test -z "$PRUNEREGEX" &&
PRUNEREGEX=`echo $PRUNEPATHS|sed -e 's,^,\\\(^,' -e 's,$\\\)\\\|\\\(^,g' -e 's,$,$\\\),'`
# The database file to build.
: ${LOCATE_DB=/usr/local/var/locate/locatedb}
# Directory to hold intermediate files.
if test -z "$TMPDIR"; then
if test -d /var/tmp; then
: ${TMPDIR=/var/tmp}
elif test -d /usr/tmp; then
: ${TMPDIR=/usr/tmp}
else
: ${TMPDIR=/tmp}
fi
fi
export TMPDIR
# The user to search network directories as.
: ${NETUSER=daemon}
# The directory containing the subprograms.
if test -n "$LIBEXECDIR" ; then
: LIBEXECDIR already set,do nothing
else
: ${LIBEXECDIR=/usr/local/Cellar/findutils/4.7.0/libexec}
fi
# The directory containing find.
if test -n "$BINDIR" ; then
: BINDIR already set,do nothing
else
: ${BINDIR=/usr/local/Cellar/findutils/4.7.0/bin}
fi
# The names of the utilities to run to build the database.
: ${find:=${BINDIR}/gfind}
: ${frcode:=${LIBEXECDIR}/gfrcode}
make_tempdir () {
# This implementation is adapted from the GNU Autoconf manual.
{
tmp=`
(umask 077 && mktemp -d "$TMPDIR/updatedbXXXXXX") 2>/dev/null
` &&
test -n "$tmp" && test -d "$tmp"
} || {
# This method is less secure than mktemp -d,but it's a fallback.
#
# We use $$ as well as $RANDOM since $RANDOM may not be available.
# We also add a time-dependent suffix. This is actually somewhat
# predictable,but then so is $$. POSIX does not require date to
# support +%N.
ts=`date +%N%s || date +%s 2>/dev/null`
tmp="$TMPDIR"/updatedb"$$"-"${RANDOM:-}${ts}"
(umask 077 && mkdir "$tmp")
}
echo "$tmp"
}
checkbinary () {
if test -x "$1" ; then
: ok
else
eval echo "updatedb needs to be able to execute $1,but cannot." >&2
exit 1
fi
}
for binary in $find $frcode
do
checkbinary $binary
done
: ${PRUNEFS="
9P
NFS
afs
autofs
cifs
coda
devfs
devpts
ftpfs
iso9660
mfs
ncpfs
nfs
nfs4
proc
shfs
smbfs
sysfs
"}
if test -n "$PRUNEFS"; then
prunefs_exp=`echo $PRUNEFS |sed -e 's/\([^ ][^ ]*\)/-o -fstype \1/g' \
-e 's/-o //' -e 's/$/ -o/'`
else
prunefs_exp=''
fi
# Make and code the file list.
# Sort case insensitively for users' convenience.
rm -f $LOCATE_DB.n
trap 'rm -f $LOCATE_DB.n; exit' HUP TERM
if {
cd "$changeto"
if test -n "$SEARCHPATHS"; then
if [ "$LOCALUSER" != "" ]; then
# : A1
su $LOCALUSER `select_shell $LOCALUSER` -c \
"$find $SEARCHPATHS $FINDOPTIONS \
\\( $prunefs_exp \
-type d -regex '$PRUNEREGEX' \\) -prune -o $print_option"
else
# : A2
# ORIGINAL VERSION : sequential find
#$find $SEARCHPATHS $FINDOPTIONS \
# \( $prunefs_exp \
# -type d -regex "$PRUNEREGEX" \) -prune -o $print_option ::: /
# Parallel version 1
#parallel -j 32 $find $SEARCHPATHS $FINDOPTIONS \
# \( $prunefs_exp \
# -type d -regex "$PRUNEREGEX" \) -prune -o $print_option ::: /
# Parallel version 2
parallel -j 32 $find {} $FINDOPTIONS \
$prunefs_exp -type d -regex $PRUNEREGEX -prune -o $print_option ::: */*
fi
fi
if test -n "$NETPATHS"; then
myuid=`getuid`
if [ "$myuid" = 0 ]; then
# : A3
su $NETUSER `select_shell $NETUSER` -c \
"$find $NETPATHS $FINDOPTIONS \\( -type d -regex '$PRUNEREGEX' -prune \\) -o $print_option" ||
exit $?
else
# : A4
$find $NETPATHS $FINDOPTIONS \( -type d -regex "$PRUNEREGEX" -prune \) -o $print_option ||
exit $?
fi
fi
} | $sort | $frcode $frcode_options > $LOCATE_DB.n
then
: OK so far
true
else
rv=$?
echo "Failed to generate $LOCATE_DB.n" >&2
rm -f $LOCATE_DB.n
exit $rv
fi
# To avoid breaking locate while this script is running,put the
# results in a temp file,then rename it atomically.
if test -s $LOCATE_DB.n; then
chmod 644 ${LOCATE_DB}.n
mv ${LOCATE_DB}.n $LOCATE_DB
else
echo "updatedb: new database would be empty" >&2
rm -f $LOCATE_DB.n
fi
exit 0
我像这样启动gupdatedb
命令:
sudo gupdatedb --prunepaths='/private/tmp /private/var/folders /private/var/tmp */Backups.backupdb /System /Volumes' --localpaths='/' --output=$HOME/locatedb_gupdatedb_ParaLLEL
更新4:
我的赏金明天到期。使用默认的gupdatedb
,所有索引编制大约需要30分钟。如果我可以使用parallel
脚本的核心正确使用gupdatedb
,即当后者使用gfind
命令进行索引时,我可以期望哪个增益因子?
/bin/sh: -c: line 0: Syntax error near unexpected token `('
/bin/sh: -c: line 0: `/usr/local/Cellar/findutils/4.7.0/bin/gfind / / ( -fstype 9P -o -fstype NFS -o -fstype afs -o -fstype autofs -o -fstype cifs -o -fstype coda -o -fstype devfs -o -fstype devpts -o -fstype ftpfs -o -fstype iso9660 -o -fstype mfs -o -fstype ncpfs -o -fstype nfs -o -fstype nfs4 -o -fstype proc -o -fstype shfs -o -fstype smbfs -o -fstype sysfs -o -type d -regex \(^/private/tmp$\)\|\(^/private/var/folders$\)\|\(^/private/var/tmp$\)\|\(^*/Backups.backupdb$\)\|\(^/System$\)\|\(^/Volumes$\) ) -prune -o -print0'
使用命令:
parallel -j32 $find {} $FINDOPTIONS \
\( $prunefs_exp \
-type d -regex "$PRUNEREGEX" \) -prune -o $print_option ::: /
?
解决方法
如果没有任何内容,则不需要:::
,如果没有任何来源,{}
也毫无意义。如果没有更多关于您真正想要并行化的内容的信息,我们就无法真正告诉您应该使用什么。
但是,例如,如果要在find
,/etc
,/usr
和/bin
中分别运行一个/opt
,则看起来像
parallel find {} -options ::: /etc /usr /bin /opt
这可以等效地表示为没有:::
:
printf '%s\n' /etc /usr /bin /opt |
parallel find {} -options
所以:::
的目的基本上是说“我想指定要在命令行上并行化的内容,而不是在标准输入上接收它们”;但是,无论哪种方式,如果您都不提供此信息,parallel
将不知道将{}
替换为什么。
我并不是说这种特殊用法对您的用例有意义,只是希望澄清文档(again)。
,要从并行使用中获得任何有意义的加速,您需要确保有足够的资源来使过程更快。这里有两个挑战:
- updatedb进程受IO约束。通常,您使用并行来利用多核系统的优势,并将CPU绑定进程分散到多个核上。
- updatedb进程需要对数据库的独占访问权(通常在/var/lib/mlcoate/mlocate.db中)。即使将updateb拆分到多个内核中可以获得任何好处,您也必须将输出放入多个数据库中。这种方法将需要传递所有数据库名称(以“:”分隔,以“ -d”定位)
除非您的系统具有多个磁盘驱动器(或正在访问网络驱动器),否则运行并行查找将不会带来什么好处。
如果您的系统具有多个磁盘驱动器(和/或网络驱动器),则可以使用类似的脚本来并行运行每个文件系统
假设您在/ mnt / disk1,/ mnt / disk2上另外装有2个磁盘
# Index root
updatedb --output=/var/lib/mlocate/local.db -E '/mnt/disk1 /mnt/disk2' &
# Index 1st extra disk (or network drive)
updatedb --output=/var/lib/mlocate/disk1.db -U /mnt/disk1 &
# Index 2nd extra disk (or network drive)
updatedb --output=/var/lib/mlocate/disk2.db -U /mnt/disk2 &
wait
您应该将环境变量LOCATE_PATH设置为指向所有数据库 导出
LOCATE_PATH=/var/lib/mlocate/local.db:/var/lib/mlocate/disk1.db:/var/lib/mlocate/disk2.db
locate ...