I'm very new to shell scripting. I recently covered a tutorial of using bash with grep.
About 5 minutes ago there was, as it happened, a practical application for this function as my browser appeared to have crashed. I wanted to, for practice, use the shell to find the process and then kill it.
I typed ps -ax | grep Chrome
But many lines appeared with lots of numbers on the left (are these "job numbers" or "process numbers"?)
Is there a way to Kill all "jobs" where Chrome exists in the name of the process? Rather than kill each one individually?
I do not recommend using killall.
And you should NOT use -9 at first.
I recommend a simple pkill:
pkill Chrome
You can use killall:
killall Chrome
Edit: Removed "-9" since it should be used as a last resource (was: killall -9 Chrome).
I like my solution better. Because adds a kind of wildcard to it. I did this:
mkdir ~/.bin
echo "ps -ef | grep -i $1 | grep -v grep | awk '{print $2}' | xargs kill -9" > ~/.bin/k
echo "alias k='/Users/YOUR_USER_HERE/.bin/k $1'" >> ~/.profile
chmod +x ~/.bin/k
Now to kill stuff you type:
k Chrome
or
k chrome
or
k chr
Just be careful with system processes or not to be killed processes.
Like if you have Calculator running and another one called Tabulator also running. If you type
k ulator it will kill'em both.
Also remember that this will kill ALL instances of the program(s) that match the variable after the command k.
Because of this, you can create two aliases inside your profile file and make 2 files inside .bin one with the command grep -i and the other without it. The one without the -i flag will be case and name sensitive.
Related
I am having some strange rebooting issues and I think it is due to an error in my shell script.
#!/bin/bash
if ps -a | grep -q gridcoin
then
echo nothing
else
sudo reboot -h now
fi
The script is supposed to work like this:
Run the ps -a command so that all processes are listed. Pipe the results to grep and have grep check to see if any of the processes have the word "gridcoin" in them. If gridcoin is running, do nothing. If gridcoin is not one of the running processes, reboot the system.
I have a cron job that runs this script once every five minutes; however, my system keeps rebooting about every five minutes even when I know for a fact that gridcoin is running.
Please take a look at the code and see if it looks right before I start trying to herd other code cats.
Respectfully,
chadrick
I see at least two problems here. First, ps -a will not show processes that don't have a controlling terminal (so, basically, non-interactive processes); you want ps -ax to show all processes.
Second, depending on the timing, the grep -q gridcoin command may be listed as a running process, and of course it finds itself, meaning that it mistakes itself for the gridcoin process. If you have the pgrep program available, replace both ps and grep with it, since it automatically avoids listing itself. Unfortunately, it doesn't have a -q option like grep, so you have to redirect its output to /dev/null:
if pgrep gridcoin >/dev/null
If pgrep is not available, there are two standardish idioms to prevent grep from finding itself:
if ps -ax | grep -q '[g]ridcoin' # Adding brackets prevents it from matching itself
if ps -ax | grep gridcoin | grep -vq grep # Remove grep processes from consideration
Also, as David C. Rankin pointed out, rebooting the entire system to restart one process is overkill; just restart that one process.
I have a Docker container running Red Hat 6.8 and I have several Java processes running. On other machines with the same OS, I have used a command similar to the following to find and kill all Java processes:
ps -ef | grep 'java' | grep -v 'grep' | awk '{print $2}' | xargs kill -9
However, on this machine, it gives me the following error:
xargs: kill: No such file or directory
Taking off the "| xargs kill -9" does work and shows me the PIDs of the processes I want to kill but for some reason, the command will not work all together.
Does anyone have any ideas why?
The immediate problem is that xargs can't find the kill command. It needs to be in your PATH, probably in /bin and/or /usr/bin. You wouldn't notice it was missing when you run the kill command directly from the shell, because most shells have a kill builtin.
Also, I agree with the comments by other users about the overall idea. There are less hacky ways to do this (killall, pkill, anything that doesn't involve grep that's relying on luck to avoid matching the wrong part of the `ps output...)
That's the top result in Google about this specific issue, if not the only one, so I just wanted to note it here after two years: kill is bundled in util-linux package (in RHEL) or procps package (in Debian) along with bunch of other utilities. Installing that package in the container solves the issue.
I would to know if it's possible to detect if amarok is playing music or not, in bash.
I have tested
#! /bin/bash
if [ "$(pidof amarok --play)" ]
then
echo amarok is playing music!
else
echo amarok is not playing!
fi
But I have this error:
pidof: invalid options on command line!
pidof thinks that --play is its option, and that is of course an error.
It would be better to check if --play is inside the commandline that was used to start amarok:
grep -- --play /proc/$(pidof amarok)/cmdline
Another option is to grep output of ps aux:
ps aux | grep [a]marok.*--play
In the both cases we suppose that there only one amarok running in the system.
Update 1.
To check if any music is playing at the moment,
you can so:
grep RUNNING /proc/asound/card*/pcm*/sub*/status
This command ps -ef | grep php returns a list of processes
I want to kill in one command or with a shell script all those processes
Thanks
The easiest way to kill all commands with a given name is to use killall:
killall php
Note, this only sends an interrupt signal. This should be enough if the processes are behaving. If they're not dying from that, you can forcibly kill them using
killall -9 php
The normal way to do this is to use xargs as in ps -ef | grep php | xargs kill, but there are several ways to do this.
ps -ef lists all processes and then you use grep to pick a few lines that mention "php". This means that also commands that have "php" as part of their command line will match, and be killed. If you really want to match the command (and not the arguments as well), it is probably better to use pgrep php.
You can use a shell backtick to provide the output of a command as arguments to another command, as in
kill `pgrep php`
If you want to kill processes only, there is a command pkill that matches a pattern to the command. This can not be used if you want to do something else with the processes though. This means that if you want to kill all processes where the command contain "php", you can do this using pkill php.
Hope this helps.
You can find its pid (it's on the first column ps prints) and use the kill command to forcibly kill it:
kill -9 <pid you found>
Use xargs:
ps -ef | grep php | grep -v grep | awk '{print $2}' | xargs kill -9
grep -v grep is exclude the command itself and awk gives the list of PIDs which are then passed kill command.
Use pkill php. More on this topic in this similar question: How can I kill a process by name instead of PID?
How do you check if a process on Mac OS X is running using the process's name in a Bash script?
I am trying to write a Bash script that will restart a process if it has stopped but do nothing if it is still running.
Parsing this:
ps aux | grep -v grep | grep -c [-i] $ProcessName
...is probably your best bet.
ps aux lists all the currently running processes including the Bash script itself which is parsed out by grep -v grep with advice from Jacob (in comments) and grep -c [-i] $ProcessName returns the optionally case-insensitive integer number of processes with integer return suggested by Sebastian.
Here's a short script that does what you're after:
#!/bin/bash
PROCESS=myapp
number=$(ps aux | grep -v grep | grep -ci $PROCESS)
if [ $number -gt 0 ]
then
echo Running;
fi
EDIT: I initially included a -i flag to grep to make it case insensitive; I did this because the example program I tried was python, which on Mac OS X runs as Python -- if you know your application's case exactly, the -i is not necessary.
The advantage of this approach is that it scales with you -- in the future, if you need to make sure, say, five instances of your application are running, you're already counting. The only caveat is if another application has your program's name in its command line, it might come up -- regular expressions to grep will resolve that issue, if you're crafty (and run into this).
Research the Darwin man pages for ps, grep, and wc.
A shorter solution:
if pgrep $PROCESS_NAME; then
echo 'Running';
fi
Explanation:
pgrep exits with 0 if there is a process matching $PROCESS_NAME running, otherwise it exist with 1.
if checks the exit code of pgrep, and, as far as exit codes go, 0 is success.
Another way is to use (abuse?) the -d option of the killall command. The -d options won't actually kill the process, but instead print what will be done. It will also exit with status 0 if it finds a matching process, or 1 if it does not. Putting this together:
#!/bin/bash
`/usr/bin/killall -d "$1" &> /dev/null`
let "RUNNING = ! $?" # this simply does a boolean 'not' on the return code
echo $RUNNING
To give credit where its due, I originally pulled this technique from a script in the iTunes installer.
This simple command will do the trick. The brackets around the process name prevent the grep command from showing in the process list. Note there is no space after the comma. There may be some portability issues as ps on some unix systems may require a dash before the options:
ps axo pid,command | grep "[S]kype"
The advantage is that you can use the results in an if statement like this:'
if [[ ! $(ps axo pid,command | grep "[i]Tunes.app") ]]; then
open -a iTunes
fi
Or if you prefer this style:
[[ ! $(ps axo pid,command | grep "[S]kype") ]] && open -a Skype || echo "Skype is up"
Another advantage is that you can get the pid by adding a pipe to awk '{print $1}'.
echo "iTunes pid: $(ps axo pid,command | grep "[i]Tunes.app" | awk '{print $1}')"
You can use either killall or kill, depending on if you are trying to find the task by PID or by name.
By Name:
if ! killall -s -0 $PROCESS_NAME >/dev/null 2>&1; then
# Restart failed app, or do whatever you need to prepare for starting the app.
else
at -f $0 +30seconds # If you don't have this on cron, you can use /usr/bin/at
fi
By PID:
if ! kill -0 $PID 2>/dev/null; then
# Restart app, do the needful.
else
at -f $0 +30seconds
fi
If you look at the OSX Manual you will see a different set of process management commands; since it's not the linux kernel, it makes sense that they would manage processes differently.
https://developer.apple.com/library/mac/documentation/Darwin/Reference/ManPages/man1/killall.1.html
A sample output from my terminal (striking out the user and hostname, of course):
user#localhost:~$ kill -0 782 # This was my old, stale SSH Agent.
bash: kill: (782) - No such process
user#localhost:~$ echo $?
1
user#localhost:~$ kill -0 813 # This is my new SSH agent, I only just created.
user#localhost:~$ echo $?
0
The return code from a kill -0 will always result in a safe way to check if the process is running, because -0 sends no signal that will ever be handled by an application. It won't kill the application, and "kill" is only called "kill" because it's usually used to stop an application.
When you look at the interfaces it uses in the source, you'll see that it's actually interacting with the process table directly (and not grepping a potentially loaded output from ps), and just sending a signal to an application. Some signals indicate the application should shutdown or stop, while other signals tell it to restart services, or re-read configuration, or re-open file descriptors to log files that have been recently rotated. There are a plethora of things that "kill" and "killall" can do that doesn't terminate the application, and it's used regularly to simply send a signal to the application.
I lack the reputation to comment on the killall answer above, but there's killall -s for doing it without sending any signals:
killall -s "$PROCESSNAME" &> /dev/null
if [ $? -eq 0 ]; then
echo "$PROCESSNAME is running"
# if you also need the PID:
PID=`killall -s "$PROCESSNAME" | awk '{print $3}'`
echo "it's PID is $PID"
fi
It has for sure!
pgrep, pkill and pfind for OpenBSD and Darwin (Mac OS X)
http://proctools.sourceforge.net
(also available via MacPorts: port info proctools )
pidof by nightproductions.net
I've extended a pidof script found on the net to use regular expressions (usually substrings) and be case insensitive
#!/bin/sh
ps axc |awk "BEGIN{ n=tolower(\"$1\")}\
tolower(\$5) ~n {print \$1}";
just create a script named "pidof" with this content, and put it in you path, i.e. in one of the dirs in
echo $PATH
and make it executable (maybe using sudo)
chmod 755 /usr/local/bin/pidof
and use it like this, of course
kill -9 `pidof pyth`
does Mac have pidof? ...
if pidof $processname >/dev/null ; then echo $processname is running ; fi
Perhaps too late for the OP but this may help others who find this thread.
The following modification of the amrox theme above works well for restarting applications on my OS X:
killall -d TextEdit &> /dev/null && killall TextEdit &> /dev/null; open -a TextEdit
I use the following AppleScript to update and restart daemons:
tell application "System Events" to set pwd to POSIX path of container of (path to me)
do shell script "launchctl unload -w /Library/LaunchDaemons/time-test.plist; cp -f " & quoted form of pwd & "/time-test.plist /Library/LaunchDaemons; launchctl load -w /Library/LaunchDaemons/time-test.plist" with administrator privileges
It assumes the original or updated plist file is in the same directory as the AppleScript.