Ssh – Using nohup to keep a script running indefinitely

cronnohupPHPrhelssh

I am very new to Linux and I've only been using it to SSH to my web server using Putty. The issue I am having is this:

Basically I've got 2 php scripts. One get_tweets.php constantly listens for tweets on twitters Streaming API whilst the other – parse_tweets.php logs these into a db. In order for these to work they've to be kept running continuously.

I've been using the following commands to run them in the background and from what I've seen they run for most of the day, however whenever I log on to my computer in the morning the scripts have stopped and I've to run commands again.

nohup php my_script.php > /dev/null &

I'm just wondering if that is normal based on the commands I am using, should they run indefinitely when using nohup and if not what is the alternative? A CRON job?

Thanks for the help

Best Answer

With nohup you must redirect also errors. The next command run script with output and errors redirecting to /dev/null:

nohup php my_script.php >/dev/null 2>&1 &

But your script can be terminated by some other reasons (error in script, oom-killer, etc). So, you should daemonise it by system's init (if it support auto-restart - upstart, systemd, and some others can do it). Or you should write cron task for check and restart your script if it not run.

Related Solutions

Execute Nohup command with input

If you just want to write a single y to the stdin of the process, you can do this:

(echo y | nohup myprocess) &

If you want to keep writing y for every prompt that comes up, the coreutil yes exists for exactly this purpose -- it will keep writing whatever you tell it to to stdout. Its default is to output "y", so you can just:

(yes | nohup myprocess) &

but if you need something else you can pass it as an argument

Cron job to check if PHP script is running, if not then run

I'd like to expand on Davidann's answer since you are new to the concept of a cron job. Every UNIX or Linux system has a crontab stored somewhere. The crontab is a plain text file. Consider the following: (From the Gentoo Wiki on Cron)

#Mins  Hours  Days   Months  Day of the week
10     3      1      1       *       /bin/echo "I don't really like cron"
30     16     *      1,2     *       /bin/echo "I like cron a little"
*      *      *      1-12/2  *       /bin/echo "I really like cron"

This crontab should echo "I really like cron" every minute of every hour of every day every other month. Obviously you would only do that if you really liked cron. The crontab will also echo "I like cron a little" at 16:30 every day in January and February. It will also echo "I don't really like cron" at 3:10 on the January 1st.

Being new to cron, you probably want to comment the starred columns so that you know what each column is used for. Every Cron implementation that I know of has always been this order. Now merging Davidann's answer with my commented file:

#Mins  Hours  Days   Months  Day of week
  *     *      *      *       *   lockfile -r 0 /tmp/the.lock && php parse_tweets.php; rm -f /tmp/the.lock

* * * * * lockfile -r 0 /tmp/the.lock && php get_tweets.php; rm -f /tmp/the.lock

Putting no value in each column defaults to:
Every Minute Of Every Hour Of Every Day Of Every Month All Week Long, --> Every Minute All Year Long.

As Davidann states using a lockfile ensures that only one copy of the php interpreter runs, php parse_tweets.php is the command to "run" the file, and the last bit of the line will delete the lock file to get ready for the next run. I don't like deleting a file every minute, but if this is the behavior you need, this is very acceptable. Writing and Rewriting to disk is just personal preference

Best Answer

Related Solutions

Execute Nohup command with input

Cron job to check if PHP script is running, if not then run

* * * * * lockfile -r 0 /tmp/the.lock && php get_tweets.php; rm -f /tmp/the.lock

Related Question