Quick-and-dirty way to ensure only one instance of a shell script is running at a time

Question

What's a quick-and-dirty way to make sure that only one instance of a shell script is running at a given time?

• Palec: Related @ Unix.SE: unix.stackexchange.com/questions/22044/…

Answer 1

Use flock(1) to make an exclusive scoped lock a on file descriptor. This way you can even synchronize different parts of the script.

#!/bin/bash
(
 # Wait for lock on /var/lock/.myscript.exclusivelock (fd 200) for 10 seconds
 flock -x -w 10 200 || exit 1
 # Do stuff
) 200>/var/lock/.myscript.exclusivelock

This ensures that code between ( and ) is run only by one process at a time and that the process doesn’t wait too long for a lock.

Caveat: this particular command is a part of util-linux. If you run an operating system other than Linux, it may or may not be available.

• chovy: What is the 200? It says "fd" in the manul, but I don't know what that means.
• Alex B: @chovy "file descriptor", an integer handle designating an open file.
• Dr. Jan-Philip Gehrcke: If anyone else is wondering: The syntax ( command A ) command B invokes a subshell for command A. Documented at tldp.org/LDP/abs/html/subshells.html. I am still not sure about the timing of invocation of the subshell and command B.
• Jonathan Leffler: I think that the code inside the sub-shell should be more like: if flock -x -w 10 200; then ...Do stuff...; else echo "Failed to lock file" 1>&2; fi so that if the timeout occurs (some other process has the file locked), this script does not go ahead and modify the file. Probably...the counter-argument is 'but if it has taken 10 seconds and the lock is still not available, it is never going to be available', presumably because the process holding the lock is not terminating (maybe it is being run under a debugger?).
• Lucas Pottersky: is "200" special? or could it be any number? i see 200s in every example

Answer 2

Naive approaches that test the existence of "lock files" are flawed.

Why? Because they don't check whether the file exists and create it in a single atomic action. Because of this; there is a race condition that WILL make your attempts at mutual exclusion break.

Instead, you can use mkdir. mkdir creates a directory if it doesn't exist yet, and if it does, it sets an exit code. More importantly, it does all this in a single atomic action making it perfect for this scenario.

if ! mkdir /tmp/myscript.lock 2>/dev/null; then
 echo "Myscript is already running." >&2
 exit 1
fi

For all details, see the excellent BashFAQ: http://mywiki.wooledge.org/BashFAQ/045

If you want to take care of stale locks, fuser(1) comes in handy. The only downside here is that the operation takes about a second, so it isn't instant.

Here's a function I wrote once that solves the problem using fuser:

# mutex file
#
# Open a mutual exclusion lock on the file, unless another process already owns one.
#
# If the file is already locked by another process, the operation fails.
# This function defines a lock on a file as having a file descriptor open to the file.
# This function uses FD 9 to open a lock on the file. To release the lock, close FD 9:
# exec 9>&-
#
mutex() {
 local file=1ドル pid pids 
 exec 9>>"$file"
 { pids=$(fuser -f "$file"); } 2>&- 9>&- 
 for pid in $pids; do
 [[ $pid = $$ ]] && continue
 exec 9>&- 
 return 1 # Locked by a pid.
 done 
}

You can use it in a script like so:

mutex /var/run/myscript.lock || { echo "Already running." >&2; exit 1; }

If you don't care about portability (these solutions should work on pretty much any UNIX box), Linux' fuser(1) offers some additional options and there is also flock(1).

• Tobias Kienzler: You can combine the if ! mkdir part with checking whether the process with the PID stored (on sucessful startup) inside the lockdir is actually running and identical to the script for stalenes protection. This would also protect against reusing the PID after a reboot, and not even require fuser.
• lhunath: It is certainly true that mkdir is not defined to be an atomic operation and as such that "side-effect" is an implementation detail of the file system. I fully believe him if he says NFS doesn't implement it in an atomic fashion. Though I don't suspect your /tmp will be an NFS share and will likely be provided by an fs that implements mkdir atomically.
• Juan Cespedes: But there is a way to check for the existence of a regular file and create it atomically if it does not: using ln to create a hard link from another file. If you have strange filesystems which don't guarantee that, you can check the inode of the new file afterwards to see if it is the same as the original file.
• Toby Speight: There is 'a way to check whether a file exists and create it in a single atomic action' - it's open(... O_CREAT|O_EXCL). You just need an suitable user program to do so, such as lockfile-create (in lockfile-progs) or dotlockfile (in liblockfile-bin). And make sure you clean up properly (e.g. trap EXIT), or test for stale locks (e.g. with --use-pid).
• Craig Hicks: "All approaches that test the existence of "lock files" are flawed. Why? Because there is no way to check whether a file exists and create it in a single atomic action. " -- To make it atomic it has to be done at the kernel level - and it is done at the kernel level with flock(1) linux.die.net/man/1/flock which appears from the man copyright date to have around since at least 2006. So I made a downvote (-1), nothing personal, just have strong conviction that using the kernel implemented tools provided by the kernel developers is correct.

Answer 3

Here's an implementation that uses a lockfile and echoes a PID into it. This serves as a protection if the process is killed before removing the pidfile:

LOCKFILE=/tmp/lock.txt
if [ -e ${LOCKFILE} ] && kill -0 `cat ${LOCKFILE}`; then
 echo "already running"
 exit
fi
# make sure the lockfile is removed when we exit and then claim it
trap "rm -f ${LOCKFILE}; exit" INT TERM EXIT
echo $$ > ${LOCKFILE}
# do stuff
sleep 1000
rm -f ${LOCKFILE}

The trick here is the kill -0 which doesn't deliver any signal but just checks if a process with the given PID exists. Also the call to trap will ensure that the lockfile is removed even when your process is killed (except kill -9).

• Paul Tomblin: As already mentioned in a comment on anther answer, this has a fatal flaw - if the other script starts up between the check and the echo, you're toast.
• bmdhacks: The symlink trick is neat, but if the owner of the lockfile is kill -9'd or the system crashes, there's still a race condition to read the symlink, notice the owner is gone, and then delete it. I'm sticking with my solution.
• dmckee --- ex-moderator kitten: Atomic check and create is available in the shell using either flock (1) or lockfile (1). See other answers.
• lhunath: See my reply for a portable way of doing an atomic check and create without having to rely on utilities such as flock or lockfile.
• K Richard Pixley: This isn't atomic and is thus useless. You need an atomic mechanism for test & set.

Answer 4

There's a wrapper around the flock(2) system call called, unimaginatively, flock(1). This makes it relatively easy to reliably obtain exclusive locks without worrying about cleanup etc. There are examples on the man page as to how to use it in a shell script.

• maxschlepzig: The flock() system call is not POSIX and does not work for files on NFS mounts.
• Ryall: Running from a Cron job I use flock -x -n %lock file% -c "%command%" to make sure only one instance is ever executing.
• Kent Kruckeberg: Aww, instead of the unimaginative flock(1) they should have went with something like flock(U). .. .it has some familiarity to it. . .seems like I've heard that before a time or two.
• Craig Hicks: It is notable that flock(2) documentation specifies use only with files, but flock(1) documentation specifies use with either file or directory. The flock(1) documentation is not explicit about how to indicate the difference during creation, but I assume it is done by adding a final "/". Anyway, if flock(1) can handle directories but flock(2) cannot, then flock(1) is not implemented only upon flock(2).

Answer 5

To make locking reliable you need an atomic operation. Many of the above proposals are not atomic. The proposed lockfile(1) utility looks promising as the man-page mentioned, that its "NFS-resistant". If your OS does not support lockfile(1) and your solution has to work on NFS, you have not many options....

NFSv2 has two atomic operations:

• symlink
• rename

With NFSv3 the create call is also atomic.

Directory operations are NOT atomic under NFSv2 and NFSv3 (please refer to the book 'NFS Illustrated' by Brent Callaghan, ISBN 0-201-32570-5; Brent is a NFS-veteran at Sun).

Knowing this, you can implement spin-locks for files and directories (in shell, not PHP):

lock current dir:

while ! ln -s . lock; do :; done

lock a file:

while ! ln -s ${f} ${f}.lock; do :; done

unlock current dir (assumption, the running process really acquired the lock):

mv lock deleteme && rm deleteme

unlock a file (assumption, the running process really acquired the lock):

mv ${f}.lock ${f}.deleteme && rm ${f}.deleteme

Remove is also not atomic, therefore first the rename (which is atomic) and then the remove.

For the symlink and rename calls, both filenames have to reside on the same filesystem. My proposal: use only simple filenames (no paths) and put file and lock into the same directory.

• maxschlepzig: Which pages of NFS Illustrated support the statement that mkdir is not atomic over NFS?
• Offirmo: Thnks for this technique. A shell mutex implementation is available in my new shell lib : github.com/Offirmo/offirmo-shell-lib, see "mutex". It uses lockfile if available, or fallback to this symlink method if not.
• Richard Hansen: Nice. Unfortunately this method does not provide a way to automatically delete stale locks.
• Matt Wallis: For the two stage unlock (mv, rm), should rm -f be used, rather than rm in case two processes P1, P2 are racing? For example, P1 commences unlock with mv, then P2 locks, then P2 unlocks (both mv and rm), finally P1 attempts rm and fails.
• Stefan Majewsky: @MattWallis That last problem could easily be mitigated by including $$ in the ${f}.deleteme filename.

Answer 6

You need an atomic operation, like flock, else this will eventually fail.

But what to do if flock is not available. Well there is mkdir. That's an atomic operation too. Only one process will result in a successful mkdir, all others will fail.

So the code is:

if mkdir /var/lock/.myscript.exclusivelock
then
 # do stuff
 :
 rmdir /var/lock/.myscript.exclusivelock
fi

You need to take care of stale locks else aftr a crash your script will never run again.

• Nippysaurus: Run this a few times concurrently (like "./a.sh & ./a.sh & ./a.sh & ./a.sh & ./a.sh & ./a.sh & ./a.sh &") and the script will leak through a few times.
• Sir Athos: @Nippysaurus: This locking method doesn't leak. What you saw was the initial script terminating before all the copies were launched, so another one was able to (correctly) get the lock. To avoid this false positive, add a sleep 10 before rmdir and try to cascade again - nothing will "leak".
• akostadinov: Other sources claim mkdir is not atomic on some filesystems like NFS. And btw I've seen occasions where on NFS concurrent recursive mkdir leads to errors sometimes with jenkins matrix jobs. So I'm pretty sure that is the case. But mkdir is pretty nice for less demanding use cases IMO.
• Palec: You can use Bash’es noclobber option with regular files.

Answer 7

You can use GNU Parallel for this as it works as a mutex when called as sem. So, in concrete terms, you can use:

sem --id SCRIPTSINGLETON yourScript

If you want a timeout too, use:

sem --id SCRIPTSINGLETON --semaphoretimeout -10 yourScript

Timeout of <0 means exit without running script if semaphore is not released within the timeout, timeout of>0 mean run the script anyway.

Note that you should give it a name (with --id) else it defaults to the controlling terminal.

GNU Parallel is a very simple install on most Linux/OSX/Unix platforms - it is just a Perl script.

• Dmitry Grigoryev: Too bad people are reluctant to downvote useless answers: this leads to new relevant answers being buried in a pile of junk.
• Partly Cloudy: We just need lots of upvotes. This such a tidy and little known answer. (Though to be pedantic OP wanted quick-and-dirty whereas this is quick-and-clean!) More on sem at related question unix.stackexchange.com/a/322200/199525 .

Answer 8

Another option is to use shell's noclobber option by running set -C. Then > will fail if the file already exists.

In brief:

set -C
lockfile="/tmp/locktest.lock"
if echo "$$" > "$lockfile"; then
 echo "Successfully acquired lock"
 # do work
 rm "$lockfile" # XXX or via trap - see below
else
 echo "Cannot acquire lock - already locked by $(cat "$lockfile")"
fi

This causes the shell to call:

open(pathname, O_CREAT|O_EXCL)

which atomically creates the file or fails if the file already exists.

---

According to a comment on BashFAQ 045, this may fail in ksh88, but it works in all my shells:

$ strace -e trace=creat,open -f /bin/bash /home/mikel/bin/testopen 2>&1 | grep -F testopen.lock
open("/tmp/testopen.lock", O_WRONLY|O_CREAT|O_EXCL|O_LARGEFILE, 0666) = 3
$ strace -e trace=creat,open -f /bin/zsh /home/mikel/bin/testopen 2>&1 | grep -F testopen.lock
open("/tmp/testopen.lock", O_WRONLY|O_CREAT|O_EXCL|O_NOCTTY|O_LARGEFILE, 0666) = 3
$ strace -e trace=creat,open -f /bin/pdksh /home/mikel/bin/testopen 2>&1 | grep -F testopen.lock
open("/tmp/testopen.lock", O_WRONLY|O_CREAT|O_EXCL|O_TRUNC|O_LARGEFILE, 0666) = 3
$ strace -e trace=creat,open -f /bin/dash /home/mikel/bin/testopen 2>&1 | grep -F testopen.lock
open("/tmp/testopen.lock", O_WRONLY|O_CREAT|O_EXCL|O_LARGEFILE, 0666) = 3

Interesting that pdksh adds the O_TRUNC flag, but obviously it's redundant:
either you're creating an empty file, or you're not doing anything.

---

How you do the rm depends on how you want unclean exits to be handled.

Delete on clean exit

New runs fail until the issue that caused the unclean exit to be resolved and the lockfile is manually removed.

# acquire lock
# do work (code here may call exit, etc.)
rm "$lockfile"

Delete on any exit

New runs succeed provided the script is not already running.

trap 'rm "$lockfile"' EXIT

• Madison Caldwell: Very novel approach... this appears to be one way to accomplish atomicity using a lock file rather than a lock directory.
• Kevin Seifert: Nice approach. :-) On the EXIT trap, it should restrict which process can clean up the lock file. For example: trap 'if [[ $(cat "$lockfile") == "$$" ]]; then rm "$lockfile"; fi' EXIT
• K Richard Pixley: Lock files aren't atomic over NFS. that's why people moved to using lock directories.
• AnyDev: IMO this is a good start, unfortunately at least bash manual does not state that it must open file with certain flags, only that noclobber will not overwrite existing file. How many code paths there are in bash and what any given flags might be used under different circumstances is unclear. This answer might be airtight practically right now, but there is no spec to claim this and nor commitment from maintainers to stick to this. IMO this answer should be used as the basis for creating the lock file without danger of clobbering existing lock file, then use flock or such to obtain a lock.

Answer 9

For shell scripts, I tend to go with the mkdir over flock as it makes the locks more portable.

Either way, using set -e isn't enough. That only exits the script if any command fails. Your locks will still be left behind.

For proper lock cleanup, you really should set your traps to something like this psuedo code (lifted, simplified and untested but from actively used scripts) :

#=======================================================================
# Predefined Global Variables
#=======================================================================
TMPDIR=/tmp/myapp
[[ ! -d $TMP_DIR ]] \
 && mkdir -p $TMP_DIR \
 && chmod 700 $TMPDIR
LOCK_DIR=$TMP_DIR/lock
#=======================================================================
# Functions
#=======================================================================
function mklock {
 __lockdir="$LOCK_DIR/$(date +%s.%N).$$" # Private Global. Use Epoch.Nano.PID
 # If it can create $LOCK_DIR then no other instance is running
 if $(mkdir $LOCK_DIR)
 then
 mkdir $__lockdir # create this instance's specific lock in queue
 LOCK_EXISTS=true # Global
 else
 echo "FATAL: Lock already exists. Another copy is running or manually lock clean up required."
 exit 1001 # Or work out some sleep_while_execution_lock elsewhere
 fi
}
function rmlock {
 [[ ! -d $__lockdir ]] \
 && echo "WARNING: Lock is missing. $__lockdir does not exist" \
 || rmdir $__lockdir
}
#-----------------------------------------------------------------------
# Private Signal Traps Functions {{{2
#
# DANGER: SIGKILL cannot be trapped. So, try not to `kill -9 PID` or 
# there will be *NO CLEAN UP*. You'll have to manually remove 
# any locks in place.
#-----------------------------------------------------------------------
function __sig_exit {
 # Place your clean up logic here 
 # Remove the LOCK
 [[ -n $LOCK_EXISTS ]] && rmlock
}
function __sig_int {
 echo "WARNING: SIGINT caught" 
 exit 1002
}
function __sig_quit {
 echo "SIGQUIT caught"
 exit 1003
}
function __sig_term {
 echo "WARNING: SIGTERM caught" 
 exit 1015
}
#=======================================================================
# Main
#=======================================================================
# Set TRAPs
trap __sig_exit EXIT # SIGEXIT
trap __sig_int INT # SIGINT
trap __sig_quit QUIT # SIGQUIT
trap __sig_term TERM # SIGTERM
mklock
# CODE
exit # No need for cleanup code here being in the __sig_exit trap function

Here's what will happen. All traps will produce an exit so the function __sig_exit will always happen (barring a SIGKILL) which cleans up your locks.

Note: my exit values are not low values. Why? Various batch processing systems make or have expectations of the numbers 0 through 31. Setting them to something else, I can have my scripts and batch streams react accordingly to the previous batch job or script.

• mojuba: Your script is way too verbose, could've been a lot shorter I think, but overall, yes, you have to set up traps in order to do this correctly. Also I'd add SIGHUP.
• bevada: This works well, except it seems to check for $LOCK_DIR whereas it removes $__lockdir. Maybe I should suggest when removing the lock you would do rm -r $LOCK_DIR?
• Mark Stinson: Thank you for the suggestion. The above was lifted code and placed in a psuedo code fashion so it will need tuning based on folks usage. However, I deliberately went with rmdir in my case as rmdir safely removes directories only_if they are empty. If folks are placing resources in them such as PID files, etc. they should alter their lock cleanup to the more aggressive rm -r $LOCK_DIR or even force it as necessary (as I have done too in special cases such as holding relative scratch files). Cheers.
• Gilles Quénot: Have you tested exit 1002 ?
• ACK_stoverflow: For anyone else checking this answer out over a decade later: don't use values like 1002 etc for exit codes. This may be shell-specific but in general return codes will wrap after 255. Try this code to see this in action: fun() { return 255; }; fun; echo $? vs fun() { return 256; }; fun; echo $?.

Answer 10

Quick and dirty?

This one-liner on the top of your script may work most of the time. If you want reliability don't do this, use the clean solution below. (Keeping this first answer for history's sake. 😅)

[[ $(pgrep -c "`basename \"0ドル\"`") -gt 1 ]] && exit

Quick and clean (and official)?

This one-liner to the top of your script came straight from the flock manpage:

[ "${FLOCKER}" != "0ドル" ] && exec env FLOCKER="0ドル" flock -en "0ドル" "0ドル" "$@" || :

This is useful boilerplate code for shell scripts. Put it at the top of the shell script you want to lock and it’ll automatically lock itself on the first run. If the environment variable $FLOCKER is not set to the shell script that is being run, then execute flock and grab an exclusive non-blocking lock (using the script itself as the lock file) before re-execing itself with the right arguments. It also sets the FLOCKER environment variable to the right value so it doesn’t run again.

Of course, just make sure that your script name is unique. :)

• rubo77: How do I simulate this to test it? Is there a way to start a script twice in one line and maybe get an warning, if it is already running?
• rubo77: This is not working at all! Why check -gt 2? grep doesn't always find itself in the result of ps!
• Palec: pgrep is not in POSIX. If you want to get this working portably, you need POSIX ps and process its output.
• Benjamin Peter: On OSX -c does not exist, you will have to use | wc -l. About the number comparison: -gt 1 is checked since the first instance sees itself.
• Majal: Updated the answer to the one I'm currently using. It was a long time ago to visit this again. 😅

Answer 11

Here's an approach that combines atomic directory locking with a check for stale lock via PID and restart if stale. Also, this does not rely on any bashisms.

#!/bin/dash
SCRIPTNAME=$(basename 0ドル)
LOCKDIR="/var/lock/${SCRIPTNAME}"
PIDFILE="${LOCKDIR}/pid"
if ! mkdir $LOCKDIR 2>/dev/null
then
 # lock failed, but check for stale one by checking if the PID is really existing
 PID=$(cat $PIDFILE)
 if ! kill -0 $PID 2>/dev/null
 then
 echo "Removing stale lock of nonexistent PID ${PID}">&2
 rm -rf $LOCKDIR
 echo "Restarting myself (${SCRIPTNAME})">&2
 exec "0ドル" "$@"
 fi
 echo "$SCRIPTNAME is already running, bailing out">&2
 exit 1
else
 # lock successfully acquired, save PID
 echo $$> $PIDFILE
fi
trap "rm -rf ${LOCKDIR}" QUIT INT TERM EXIT
echo hello
sleep 30s
echo bye

• MaXi32: nice readable and most importantly it has everything that the democratic people are arguing with. This is true democracy.

Answer 12

Add this line at the beginning of your script

[ "${FLOCKER}" != "0ドル" ] && exec env FLOCKER="0ドル" flock -en "0ドル" "0ドル" "$@" || :

It's a boilerplate code from man flock.

If you want more logging, use this one

[ "${FLOCKER}" != "0ドル" ] && { echo "Trying to start build from queue... "; exec bash -c "FLOCKER='0ドル' flock -E $E_LOCKED -en '0ドル' '0ドル' '$@' || if [ \"\$?\" -eq $E_LOCKED ]; then echo 'Locked.'; fi"; } || echo "Lock is free. Completing."

This sets and checks locks using flock utility. This code detects if it was run first time by checking FLOCKER variable, if it is not set to script name, then it tries to start script again recursively using flock and with FLOCKER variable initialized, if FLOCKER is set correctly, then flock on previous iteration succeeded and it is OK to proceed. If lock is busy, it fails with configurable exit code.

It seems to not work on Debian 7, but seems to work back again with experimental util-linux 2.25 package. It writes "flock: ... Text file busy". It could be overridden by disabling write permission on your script.

• Mihail Kostira: What does the last part of this snippet do: || :
• user3132194: @Mihail it means do nothing if test is false. In second example I use echo instead of colon on this place. Here are good description for colon operator stackoverflow.com/a/3224910/3132194
• Tomas: Recommend this answer because it just lie in official manual, and no extra lock file is needed!
• AnyDev: Be aware that by locking on the script file itself instead of a dedicated lock file you are risking situation where script gets replaced (updated or edited), therefore another copy successfully locks on new script file even though already running script is still locking the previous script version that was deleted. I used to run into this problem after package updates and/or script edits with vim.

Answer 13

Create a lock file in a known location and check for existence on script start? Putting the PID in the file might be helpful if someone's attempting to track down an errant instance that's preventing execution of the script.

Answer 14

This example is explained in the man flock, but it needs some impovements, because we should manage bugs and exit codes:

 #!/bin/bash
 #set -e this is useful only for very stupid scripts because script fails when anything command exits with status more than 0 !! without possibility for capture exit codes. not all commands exits >0 are failed.
( #start subprocess
 # Wait for lock on /var/lock/.myscript.exclusivelock (fd 200) for 10 seconds
 flock -x -w 10 200
 if [ "$?" != "0" ]; then echo Cannot lock!; exit 1; fi
 echo $$>>/var/lock/.myscript.exclusivelock #for backward lockdir compatibility, notice this command is executed AFTER command bottom ) 200>/var/lock/.myscript.exclusivelock.
 # Do stuff
 # you can properly manage exit codes with multiple command and process algorithm.
 # I suggest throw this all to external procedure than can properly handle exit X commands
) 200>/var/lock/.myscript.exclusivelock #exit subprocess
FLOCKEXIT=$? #save exitcode status
 #do some finish commands
exit $FLOCKEXIT #return properly exitcode, may be usefull inside external scripts

You can use another method, list processes that I used in the past. But this is more complicated that method above. You should list processes by ps, filter by its name, additional filter grep -v grep for remove parasite nad finally count it by grep -c . and compare with number. Its complicated and uncertain

• Znik: You can use ln -s , because this can create symlink only when no file or symlink exists, the same as mkdir. a lot of system processes used symlinks in the past, for example init or inetd. synlink keeps process id, but really points to nothing. for the years this behavior was changed. processes uses flocks and semaphores.

Answer 15

The existing answers posted either rely on the CLI utility flock or do not properly secure the lock file. The flock utility is not available on all non-Linux systems (i.e. FreeBSD), and does not work properly on NFS.

In my early days of system administration and system development, I was told that a safe and relatively portable method of creating a lock file was to create a temp file using mkemp(3) or mkemp(1), write identifying information to the temp file (i.e. PID), then hard link the temp file to the lock file. If the link was successful, then you have successfully obtained the lock.

When using locks in shell scripts, I typically place an obtain_lock() function in a shared profile and then source it from the scripts. Below is an example of my lock function:

obtain_lock()
{
 LOCK="${1}"
 LOCKDIR="$(dirname "${LOCK}")"
 LOCKFILE="$(basename "${LOCK}")"
 # create temp lock file
 TMPLOCK=$(mktemp -p "${LOCKDIR}" "${LOCKFILE}XXXXXX" 2> /dev/null)
 if test "x${TMPLOCK}" == "x";then
 echo "unable to create temporary file with mktemp" 1>&2
 return 1
 fi
 echo "$$" > "${TMPLOCK}"
 # attempt to obtain lock file
 ln "${TMPLOCK}" "${LOCK}" 2> /dev/null
 if test $? -ne 0;then
 rm -f "${TMPLOCK}"
 echo "unable to obtain lockfile" 1>&2
 if test -f "${LOCK}";then
 echo "current lock information held by: $(cat "${LOCK}")" 1>&2
 fi
 return 2
 fi
 rm -f "${TMPLOCK}"
 return 0;
};

The following is an example of how to use the lock function:

#!/bin/sh
. /path/to/locking/profile.sh
PROG_LOCKFILE="/tmp/myprog.lock"
clean_up()
{
 rm -f "${PROG_LOCKFILE}"
}
obtain_lock "${PROG_LOCKFILE}"
if test $? -ne 0;then
 exit 1
fi
trap clean_up SIGHUP SIGINT SIGTERM
# bulk of script
clean_up
exit 0
# end of script

Remember to call clean_up at any exit points in your script.

I've used the above in both Linux and FreeBSD environments.

Answer 16

If flock's limitations, which have already been described elsewhere on this thread, aren't an issue for you, then this should work:

#!/bin/bash
{
 # exit if we are unable to obtain a lock; this would happen if 
 # the script is already running elsewhere
 # note: -x (exclusive) is the default
 flock -n 100 || exit
 # put commands to run here
 sleep 100
} 100>/tmp/myjob.lock 

• Keldon Alleyne: Just thought I'd point out that -x (write lock) is already set by default.
• Anentropic: and -n will exit 1 immediately if it can't get the lock
• presto8: Thanks @KeldonAlleyne, I updated the code to remove "-x" since it is default.

Answer 17

When targeting a Debian machine I find the lockfile-progs package to be a good solution. procmail also comes with a lockfile tool. However sometimes I am stuck with neither of these.

Here's my solution which uses mkdir for atomic-ness and a PID file to detect stale locks. This code is currently in production on a Cygwin setup and works well.

To use it simply call exclusive_lock_require when you need get exclusive access to something. An optional lock name parameter lets you share locks between different scripts. There's also two lower level functions (exclusive_lock_try and exclusive_lock_retry) should you need something more complex.

function exclusive_lock_try() # [lockname]
{
 local LOCK_NAME="${1:-`basename 0ドル`}"
 LOCK_DIR="/tmp/.${LOCK_NAME}.lock"
 local LOCK_PID_FILE="${LOCK_DIR}/${LOCK_NAME}.pid"
 if [ -e "$LOCK_DIR" ]
 then
 local LOCK_PID="`cat "$LOCK_PID_FILE" 2> /dev/null`"
 if [ ! -z "$LOCK_PID" ] && kill -0 "$LOCK_PID" 2> /dev/null
 then
 # locked by non-dead process
 echo "\"$LOCK_NAME\" lock currently held by PID $LOCK_PID"
 return 1
 else
 # orphaned lock, take it over
 ( echo $$ > "$LOCK_PID_FILE" ) 2> /dev/null && local LOCK_PID="$$"
 fi
 fi
 if [ "`trap -p EXIT`" != "" ]
 then
 # already have an EXIT trap
 echo "Cannot get lock, already have an EXIT trap"
 return 1
 fi
 if [ "$LOCK_PID" != "$$" ] &&
 ! ( umask 077 && mkdir "$LOCK_DIR" && umask 177 && echo $$ > "$LOCK_PID_FILE" ) 2> /dev/null
 then
 local LOCK_PID="`cat "$LOCK_PID_FILE" 2> /dev/null`"
 # unable to acquire lock, new process got in first
 echo "\"$LOCK_NAME\" lock currently held by PID $LOCK_PID"
 return 1
 fi
 trap "/bin/rm -rf \"$LOCK_DIR\"; exit;" EXIT
 return 0 # got lock
}
function exclusive_lock_retry() # [lockname] [retries] [delay]
{
 local LOCK_NAME="1ドル"
 local MAX_TRIES="${2:-5}"
 local DELAY="${3:-2}"
 local TRIES=0
 local LOCK_RETVAL
 while [ "$TRIES" -lt "$MAX_TRIES" ]
 do
 if [ "$TRIES" -gt 0 ]
 then
 sleep "$DELAY"
 fi
 local TRIES=$(( $TRIES + 1 ))
 if [ "$TRIES" -lt "$MAX_TRIES" ]
 then
 exclusive_lock_try "$LOCK_NAME" > /dev/null
 else
 exclusive_lock_try "$LOCK_NAME"
 fi
 LOCK_RETVAL="${PIPESTATUS[0]}"
 if [ "$LOCK_RETVAL" -eq 0 ]
 then
 return 0
 fi
 done
 return "$LOCK_RETVAL"
}
function exclusive_lock_require() # [lockname] [retries] [delay]
{
 if ! exclusive_lock_retry "$@"
 then
 exit 1
 fi
}

• ndemou: Thanks, tried it on cygwin myself and it passed simple tests.

Answer 18

Some unixes have lockfile which is very similar to the already mentioned flock.

From the manpage:

lockfile can be used to create one or more semaphore files. If lock- file can't create all the specified files (in the specified order), it waits sleeptime (defaults to 8) seconds and retries the last file that didn't succeed. You can specify the number of retries to do until failure is returned. If the number of retries is -1 (default, i.e., -r-1) lockfile will retry forever.

• Offirmo: how do we get the lockfile utility ??
• Mr. Deathless: lockfile is distributed with procmail. Also there is an alternative dotlockfile that goes with liblockfile package. They both claim to work reliably on NFS.

Answer 19

I use a simple approach that handles stale lock files.

Note that some of the above solutions that store the pid, ignore the fact that the pid can wrap around. So - just checking if there is a valid process with the stored pid is not enough, especially for long running scripts.

I use noclobber to make sure only one script can open and write to the lock file at one time. Further, I store enough information to uniquely identify a process in the lockfile. I define the set of data to uniquely identify a process to be pid,ppid,lstart.

When a new script starts up, if it fails to create the lock file, it then verifies that the process that created the lock file is still around. If not, we assume the original process died an ungraceful death, and left a stale lock file. The new script then takes ownership of the lock file, and all is well the world, again.

Should work with multiple shells across multiple platforms. Fast, portable and simple.

#!/usr/bin/env sh
# Author: rouble
LOCKFILE=/var/tmp/lockfile #customize this line
trap release INT TERM EXIT
# Creates a lockfile. Sets global variable $ACQUIRED to true on success.
# 
# Returns 0 if it is successfully able to create lockfile.
acquire () {
 set -C #Shell noclobber option. If file exists, > will fail.
 UUID=`ps -eo pid,ppid,lstart $$ | tail -1`
 if (echo "$UUID" > "$LOCKFILE") 2>/dev/null; then
 ACQUIRED="TRUE"
 return 0
 else
 if [ -e $LOCKFILE ]; then 
 # We may be dealing with a stale lock file.
 # Bring out the magnifying glass. 
 CURRENT_UUID_FROM_LOCKFILE=`cat $LOCKFILE`
 CURRENT_PID_FROM_LOCKFILE=`cat $LOCKFILE | cut -f 1 -d " "`
 CURRENT_UUID_FROM_PS=`ps -eo pid,ppid,lstart $CURRENT_PID_FROM_LOCKFILE | tail -1`
 if [ "$CURRENT_UUID_FROM_LOCKFILE" == "$CURRENT_UUID_FROM_PS" ]; then 
 echo "Script already running with following identification: $CURRENT_UUID_FROM_LOCKFILE" >&2
 return 1
 else
 # The process that created this lock file died an ungraceful death. 
 # Take ownership of the lock file.
 echo "The process $CURRENT_UUID_FROM_LOCKFILE is no longer around. Taking ownership of $LOCKFILE"
 release "FORCE"
 if (echo "$UUID" > "$LOCKFILE") 2>/dev/null; then
 ACQUIRED="TRUE"
 return 0
 else
 echo "Cannot write to $LOCKFILE. Error." >&2
 return 1
 fi
 fi
 else
 echo "Do you have write permissons to $LOCKFILE ?" >&2
 return 1
 fi
 fi
}
# Removes the lock file only if this script created it ($ACQUIRED is set), 
# OR, if we are removing a stale lock file (first parameter is "FORCE") 
release () {
 #Destroy lock file. Take no prisoners.
 if [ "$ACQUIRED" ] || [ "1ドル" == "FORCE" ]; then
 rm -f $LOCKFILE
 fi
}
# Test code
# int main( int argc, const char* argv[] )
echo "Acquring lock."
acquire
if [ $? -eq 0 ]; then 
 echo "Acquired lock."
 read -p "Press [Enter] key to release lock..."
 release
 echo "Released lock."
else
 echo "Unable to acquire lock."
fi

• Tagar: I gave you +1 for a different solution. Althoug it doesn't work neither in AIX (> ps -eo pid,ppid,lstart $$ | tail -1 ps: invalid list with -o.) not HP-UX (> ps -eo pid,ppid,lstart $$ | tail -1 ps: illegal option -- o). Thanks.

Answer 20

I wanted to do away with lockfiles, lockdirs, special locking programs and even pidof since it isn't found on all Linux installations. Also wanted to have the simplest code possible (or at least as few lines as possible). Simplest if statement, in one line:

if [[ $(ps axf | awk -v pid=$$ '1ドル!=pid && 6ドル~/'$(basename 0ドル)'/{print 1ドル}') ]]; then echo "Already running"; exit; fi

• qneill: This is sensitive to the 'ps' output, on my machine (Ubuntu 14.04, /bin/ps from procps-ng version 3.3.9) the 'ps axf' command prints ascii tree characters which disrupt the field numbers. This worked for me: /bin/ps -a --format pid,cmd | awk -v pid=$$ '/'$(basename 0ドル)'/ { if (1ドル!=pid) print 1ドル; }'

Answer 21

An example with flock(1) but without subshell. flock()ed file /tmp/foo is never removed, but that doesn't matter as it gets flock() and un-flock()ed.

#!/bin/bash
exec 9<> /tmp/foo
flock -n 9
RET=$?
if [[ $RET -ne 0 ]] ; then
 echo "lock failed, exiting"
 exit
fi
#Now we are inside the "critical section"
echo "inside lock"
sleep 5
exec 9>&- #close fd 9, and release lock
#The part below is outside the critical section (the lock)
echo "lock released"
sleep 5

• Wolfson: This is what I use, except that I put the lock-check into a while loop: while ! flock -n 9; do sleep 1 done so that the other instance will continue as soon as the lock is removed.

Answer 22

Actually although the answer of bmdhacks is almost good, there is a slight chance the second script to run after first checked the lockfile and before it wrote it. So they both will write the lock file and they will both be running. Here is how to make it work for sure:

lockfile=/var/lock/myscript.lock
if ( set -o noclobber; echo "$$" > "$lockfile") 2> /dev/null ; then
 trap 'rm -f "$lockfile"; exit $?' INT TERM EXIT
else
 # or you can decide to skip the "else" part if you want
 echo "Another instance is already running!"
fi

The noclobber option will make sure that redirect command will fail if file already exists. So the redirect command is actually atomic - you write and check the file with one command. You don't need to remove the lockfile at the end of file - it'll be removed by the trap. I hope this helps to people that will read it later.

P.S. I didn't see that Mikel already answered the question correctly, although he didn't include the trap command to reduce the chance the lock file will be left over after stopping the script with Ctrl-C for example. So this is the complete solution

Answer 23

This one line answer comes from someone related Ask Ubuntu Q&A:

[ "${FLOCKER}" != "0ドル" ] && exec env FLOCKER="0ドル" flock -en "0ドル" "0ドル" "$@" || :
# This is useful boilerplate code for shell scripts. Put it at the top of
# the shell script you want to lock and it'll automatically lock itself on
# the first run. If the env var $FLOCKER is not set to the shell script
# that is being run, then execute flock and grab an exclusive non-blocking
# lock (using the script itself as the lock file) before re-execing itself
# with the right arguments. It also sets the FLOCKER env var to the right
# value so it doesn't run again.

Answer 24

I have following problems with the existing answers:

• Some answers use script file itself 0ドル or $BASH_SOURCE for locking often referring to examples from man flock. This fails when the script file is replaced due to update or edit causing next run to open and obtain a lock on the new script file even though another instance holding a lock on the removed file is still running.
• Few answers use a fixed file descriptor. This is not ideal. I do not want to rely on how this will behave e.g. opening lock file fails but gets mishandled and attempts to lock on unrelated file descriptor inherited from parent process. Another fail case is injecting locking wrapper for a 3rd party binary that does not handle locking itself but fixed file descriptors can interfere with file descriptor passing to child processes.
• I reject answers using process lookup for already running script name. There are several reasons for it, such as but not limited to reliability/atomicity, parsing output, and having script that does several related functions some of which do not require locking.
• Some answers try to clean up lock files and then having to deal with stale lock files caused by e.g. sudden crash/reboot. IMO that is unnecessarily complicated. Let lock files stay.

This answer does:

• rely on flock because it gets kernel to provide locking and therefore benefits from flock features such as timeout for attempting to acquire a lock (without polling) and shared vs exclusive lock type, something that would be messy to do by most other means.
• assume and rely on lock file being stored on the local filesystem as opposed to NFS. I assume the goal is to allow programs running on same host to lock each other out. Distributed applications are "out of scope" here.
• change lock file presence to NOT mean anything about a running instance. Its role is purely to prevent two concurrent instances creating file with same name and replacing another's copy. Lock file does not get deleted, it gets left behind and allowed to remain across reboots. The locking is indicated via flock and NOT via lock file existence.
• assume bash shell, as tagged by the question.

It's not a oneliner, but without comments nor error messages it's small enough:

#!/bin/bash
LOCKFILE=/var/lock/TODO
set -o noclobber
exec {lockfd}<> "${LOCKFILE}" || exit 1
set +o noclobber # depends on what you need
flock --exclusive --nonblock ${lockfd} || exit 1

But I prefer comments and error messages:

#!/bin/bash
# TODO Set a lock file name
LOCKFILE=/var/lock/myprogram.lock
# Set noclobber option to ensure lock file is not REPLACED.
set -o noclobber
# Open lock file for R+W on a new file descriptor
# and assign the new file descriptor to "lockfd" variable.
# If lock file does not exist it gets created non-destructively
# without loops and without race condition due to 'set -o noclobber'
# This does NOT obtain a lock.
exec {lockfd}<> "${LOCKFILE}" || {
 echo "pid=$$ failed to open LOCKFILE='${LOCKFILE}'" 1>&2
 exit 1
}
# TODO!!!! undo/set the desired noclobber value for the remainder of the script
set +o noclobber
# Lock on the allocated file descriptor or fail
# TODO: Adjust flock options e.g. --noblock or --timeout as needed
flock --exclusive --nonblock ${lockfd} || {
 echo "pid=$$ failed to obtain lock fd='${lockfd}' LOCKFILE='${LOCKFILE}'" 1>&2
 exit 1
}
# DO work here
echo "pid=$$ obtained exclusive lock fd='${lockfd}' LOCKFILE='${LOCKFILE}'"
# TODO: After critical code can unlock and do more work after unlocking
#flock -u ${lockfd};
# if unlocking then might as well close lockfd too
#exec {lockfd}<&-

• AnyDev: And no, I don't write PID to the lock file, I don't want anyone applying a habit of kill $(cat lockfile) and killing unrelated process which is a problem that would happen when relying on lock file presence and having to clean stale lock files. No cleaning required - no problem.
• Yury Ershov: What is that exec {fd} <> X syntax? It seems to work in recent versions of Bash but I can't find anything in the docs. This curly braces thing is something new.
• Yury Ershov: Ah got it. It's not exec {fd} but rather {fd}<> thing: Each redirection that may be preceded by a file descriptor number may instead be preceded by a word of the form {varname}. One comment: whenever I need a "no-op" command in Bash, I use : (not exec). Exec thing is somewhat confusing since it's meant to pass the execution to another command and never continue the current script. Your thing would then look like: : {lockfd}<> "${LOCKFILE}"
• AnyDev: @YuriyErshov I had to read up and found this in bash docs regarding file descriptor redirection: If {varname} is supplied, the redirection persists beyond the scope of the command, allowing the shell programmer to manage the file descriptor himself.. When I wrote my answer I was not aware of that the persists beyond the scope of the command part. That makes your comment about replacing exec with : very useful indeed, thanks!
• Yury Ershov: By the way, it's probably a good idea to scope the lock by running it in a subshell (( ... )) so that the locked file descriptor doesn't leak.

Answer 25

PID and lockfiles are definitely the most reliable. When you attempt to run the program, it can check for the lockfile which and if it exists, it can use ps to see if the process is still running. If it's not, the script can start, updating the PID in the lockfile to its own.

Answer 26

I find that bmdhack's solution is the most practical, at least for my use case. Using flock and lockfile rely on removing the lockfile using rm when the script terminates, which can't always be guaranteed (e.g., kill -9).

I would change one minor thing about bmdhack's solution: It makes a point of removing the lock file, without stating that this is unnecessary for the safe working of this semaphore. His use of kill -0 ensures that an old lockfile for a dead process will simply be ignored/over-written.

My simplified solution is therefore to simply add the following to the top of your singleton:

## Test the lock
LOCKFILE=/tmp/singleton.lock 
if [ -e ${LOCKFILE} ] && kill -0 `cat ${LOCKFILE}`; then
 echo "Script already running. bye!"
 exit 
fi
## Set the lock 
echo $$ > ${LOCKFILE}

Of course, this script still has the flaw that processes that are likely to start at the same time have a race hazard, as the lock test and set operations are not a single atomic action. But the proposed solution for this by lhunath to use mkdir has the flaw that a killed script may leave behind the directory, thus preventing other instances from running.

Answer 27

Quick and dirty?

#!/bin/sh
if [ -f sometempfile ]
 echo "Already running... will now terminate."
 exit
else
 touch sometempfile
fi
..do what you want here..
rm sometempfile

• TimB: This may or may not be an issue, depending on how it's used, but there's a race condition between testing for the lock and creating it, so that two scripts could both be started at the same time. If one terminates first, the other will stay running with no lock file.
• Paul Tomblin: C News, which taught me much about portable shell scripting, used to make a lock.$$ file, and then attempt to link it with "lock" - if the link succeeed, you had the lock, otherwise you removed lock.$$ and exited.
• Matthew Scharley: That's a really good way to do it, except you still suffer the need to remove the lockfile manually if something goes wrong and the lockfile isn't deleted.
• Aupajo: Quick and dirty, that's what he asked for :)

Answer 28

The semaphoric utility uses flock (as discussed above, e.g. by presto8) to implement a counting semaphore. It enables any specific number of concurrent processes you want. We use it to limit the level of concurrency of various queue worker processes.

It's like sem but much lighter-weight. (Full disclosure: I wrote it after finding the sem was way too heavy for our needs and there wasn't a simple counting semaphore utility available.)

Answer 29

Answered a million times already, but another way, without the need for external dependencies:

LOCK_FILE="/var/lock/$(basename "0ドル").pid"
trap "rm -f ${LOCK_FILE}; exit" INT TERM EXIT
if [[ -f $LOCK_FILE && -d /proc/`cat $LOCK_FILE` ]]; then
 // Process already exists
 exit 1
fi
echo $$ > $LOCK_FILE

Each time it writes the current PID ($$) into the lockfile and on script startup checks if a process is running with the latest PID.

• Philippe Chaintreuil: Without the trap call (or at least a cleanup near the end for the normal case), you have the false positive bug where the lockfile is left around after the last run and the PID has been reused by another process later. (And in the worst case, it's been gifted to a long running process like apache....)
• Filidor Wiese: I agree, my approach is flawed, it does need a trap. I've updated my solution. I still prefer to not have external dependencies.

Answer 30

Using the process's lock is much stronger and takes care of the ungraceful exits also. lock_file is kept open as long as the process is running. It will be closed (by shell) once the process exists (even if it gets killed). I found this to be very efficient:

lock_file=/tmp/`basename 0ドル`.lock
if fuser $lock_file > /dev/null 2>&1; then
 echo "WARNING: Other instance of $(basename 0ドル) running."
 exit 1
fi
exec 3> $lock_file