How to kill running processes with multiple instances

I have my .bashrc file setup with many aliases, you can see more here. One of them is to kill processes, but sometimes it's not useful for running processes with multiple instances.

My alias looks like:

alias fproces='ps x | grep '

So when I get a process or an app that I want to kill I can do the following:

$ fproces gedit 7892 ? S 0:00 gedit 7894 pts/1 R+ 0:00 grep gedit $ kill -9 7892

This is not that useful for applications/processes like Firefox:

$ fproces firefox 5994 ? S 0:00 /bin/sh /usr/bin/firefox 6006 ? S 0:00 /bin/sh /usr/lib/firefox/run-mozilla.sh /usr/lib/firefox/firefox-bin 6011 ? Sl 54:24 /usr/lib/firefox/firefox-bin 7898 pts/1 R+ 0:00 grep firefox $ kill -9 5994 6006 6011

So I came up with a little script that does the following:

$ finish.sh Enter app name firefox 5994 ? S 0:00 /bin/sh /usr/bin/firefox 6006 ? S 0:00 /bin/sh /usr/lib/firefox/run-mozilla.sh /usr/lib/firefox/firefox-bin 6011 ? Sl 54:38 /usr/lib/firefox/firefox-bin Do you want to kill all items? [y|n]

The script lists all processes related to the app and kill it if you say “y”. In this example I typed in “firefox” when prompted for an app name.

Here is what the script looks like:

#!/bin/bash ## reads process name and assign it to var app echo "Enter app name" read app ## sets a minimum size for the var app. Without this, you could easily kill all running processes minim=`echo "$app" | wc -m` if [ "$minim" -lt "4" ] ; then echo "Please try again with an app name" exit 1 fi ## looks for the running process instances, lists it and asks if you want to kill all of them ps x | grep "$app" | grep -v 'grep' echo echo "Do you want to kill all items? [y|n] " ## reads the answer and acts if it's an “y” read resposta if [ "$resposta" = "y" ] ; then termina=`ps x | grep "$app" | grep -v 'grep' | awk '{print $1 ; }'` for i in `echo $termina` ; do kill -9 "$i" done fi

Vic.

How to get email and sms notifications from your server

Ever wondered if you could make you server send you an email or a sms to your cell phone if it ever went down, or if you logs folder were too big, or any other event that you can think? Well, here's a really simple way of doing so.

In my case, I'll be using two events, one if the server goes down, and another if my logs folder gets too big (I have a very simple Linux box that does not have file rotation).


1- Mayday, server is down

There are two ways we can do this, and it will depend on how often do you need to check if the server is down. If you need it to be done every minute or so, I'd advise using a service instead of crontab. I could not personally say what's the system hog difference, but I'm open for feedbacks. In my case I'll be checking every hour, so I'll use crontab.

Other thing required is a cli email client, and there are many you can use. I choose msmtp, which is extremely easy to configure. Here's what I have set to use my gmail account:

defaults tls on tls_certcheck off logfile ~/.msmtp.log account gmail host smtp.gmail.com port 587 from me@gmail.com auth on user me@gmail.com password ###### account default : gmail

To send an email is as easy as:

$ echo -e "Subject: Test \n\n This is the message body" | msmtp -v email@address1 email@address2

So now that we have means to send the message, let's create the script. Remember that the script will need to be run from another PC and not from the server. A server that is down cannot send emails!!

#!/bin/bash ## pings the server IP. Here I'm using only one ping as some distros ping 4x times (Fedora), and others will ping consecutively by default (Ubuntu) ping -c 1 server ## checks if the ping was successful or not, if not it sends an email and a sms with the last time it tried to ping if [ “$?” != “0” ] ; then echo -e "Subject: Server is down \n\n Victor, \n Server is down. Last time tested was $(date)." | msmtp -v me@adreess.ca 12312345678@cellprovider.com fi

Now we add the script to your crontab and send any error msg to a txt. Make sure that your account has pinging privileges

$ crontab -e # Check if server is up every hour 0 * * * * /home/user/bin/script.sh 2>> /home/user/var/log/server-up-log.txt

2- Checking /var/logs size

The process here will be almost the same, however both script and crontab need to be run on the server. Our script should look something like:

#!/bin/bash ## Some txt editing. We will check /var/log size by using “df” and then striping it down to numbers only e=`df -h | grep 'var\/log' | awk '{ print $5 ;}' | sed 's/%//'` ## If the size of /var/log gets higher than 75% we will get an email and a sms if [ "$e" -gt "75" ] ; then echo -e "Subject: Log folder \n\n Victor, \n Log folder has reached 75%" | msmtp -v me@adreess.ca 12312345678@cellprovider.com fi

Add the script to your crontab on the server and you are done.

Have fun!!!


Vic.

Quick and simple Quake Terminal

For the gamers out there, you know exactly what I'm talking about. I've never been big with gaming, but the idea of having a shortcut to shade a terminal window and bring it back whenever I want is pretty neat.

I already run a small terminal window as part of my desktop, but sometimes its not as flexible as a normal window. I'll write a post on how to use a terminal as a desktop background in the future. For now you can see a pic here.

To achieve what we want we will use a utility called wmctrl, which is on the Ubuntu repos (if I'm not mistaken the Universe repo). There are other ways you can achieve the same (like Tilda).

$ sudo apt-get install wmctrl

wmctrl bases it's commands on window name, so to make sure you have the proper name for the terminal window, open it with the desired profile and type:

$ wmctrl -l 0x03800021 -1 hostname Left 0x03c000cc 0 hostname Jinzora Media Jukebox - Genres - Mozilla Firefox 0x01407084 1 hostname blog - File Browser 0x04400057 1 hostname quake-terminal - OpenOffice.org Writer 0x0380859b 1 hostname victor@hostname: ~

These are the list of windows I have open. My terminal window is called “victor@hostname”, so this is the name I'm going to use. Next, I'll open my favorite text editor and add the following lines to a script:

#!/bin/bash if [ -f ~/bin/temp/quake ] ; then wmctrl -r 'victor@hostname: ~' -b remove,below wmctrl -r 'victor@hostname: ~' -b remove,shaded rm -r ~/bin/temp/quake else wmctrl -r 'victor@hostname: ~' -b add,below wmctrl -r 'victor@hostname: ~' -b add,shaded touch ~/bin/temp/quake fi

Here's a breakdown to make it easy. I have a folder on /home/victor/bin with many scripts I created, so I'll add this one as well, and I called it ter-quake.sh. Next, I created a temp folder inside my bin folder to generate a file that will check if my script was run previously or not. And here's the explanation for the script

#!/bin/bash ## This first line checks if the file /bin/temp/quake was created, which meas the script was run before if [ -f ~/bin/temp/quake ] ; then ## if the script was run before, it will unshade the terminal window and place it on top wmctrl -r 'victor@victor-laptop: ~' -b remove,below wmctrl -r 'victor@victor-laptop: ~' -b remove,shaded ## here we remove the file rm -r ~/bin/temp/quake ## otherwise, if the file does not exist else ## the sript will shade the window, place it bellow all others and... wmctrl -r 'victor@victor-laptop: ~' -b add,below wmctrl -r 'victor@victor-laptop: ~' -b add,shaded ## create the file touch ~/bin/temp/quake fi

Save the script and make it executable:

$ sudo chmod a+x ~/bin/ter-quake.sh

Last step, we will add a keyboard shortcut for the script. I've achieved that by using gnome's gconf-editor:

- Type “Alt+F2” and enter “gconf-editor”
- Browse to “/apps/metacity/keybinding_commands”
- Choose any of the empty commands and add the path to your script
- Go to “/apps/metacity/global_keybindings” and under “run_command_#” add the shortcut you would like to use. I've used q.


Vic.


Note: This is another useful topic from Linux Journal that I wanted to share. Issue 167. Link

How to check for a running screen when logging via ssh

For some of you that might have never used screen, it's a program that allows you to create multiple virtual terminals inside one terminal window. You can even break the terminal window into two screens.

I'm not getting into details for the program's usage. You can get more information on these two links:

GNU Screen
How to split screen your command line

In my case, because screen preserves a running program even if your ssh session has lost connection, I use it to download torrents via ctorrent on one of my servers. So when I connect back to my server to check the torrent download status, I have to issue a command like the one bellow to check for running screens, and then connect to it:

$ screen -ls There is a screen on: 7583.pts-0.pluto (Detached) 1 Socket in /var/run/screen/S-webm. $ screen -r 7583

A simpler way of doing this would be:

$ screen -DD -R

This command will check for a running screen, and if none is available it will create one. So to make my live a bit easier (and my login a bit fancier), I've added a little script to my ~/.bash_profile (Ubuntu) to prompt me if I want to connect. This is what I get when I login:

Last login: xxxxxxxxxxxxxxxxxxxxxxxxxxxx There is a screen on: 7583.pts-0.pluto (Detached) 1 Socket in /var/run/screen/S-webm. A screen was detected, would you like to connect? [y|n]

To achieve that, I appended the following script to my .bash_profile:

## Check for a running Screen scrtst=`screen -ls | grep 'No Sockets' > /dev/null` if [ "$?" != "0" ] ; then screen -ls echo "A screen was detected, would you like to connect? [y|n] " read opcao if [ "$opcao" = "y" ] ; then screen -DD -R fi fi

There are many different ways you can do this. You could even check how many screens are running and list an option to connect to, but I'll leave that up to your imagination.


Vic.



Notes:
- Opcao means option in Portuguese. I like using Portuguese words for variables.
- This idea was adapted from a Linux Journal column - Tech Tip, Use Screen to Avoid Loosing Remote Work - Issue 167