问题
I'm having trouble starting an Upstart script.
Here's the script (app.conf in /etc/init/)
description "node.js server"
author "kvz"
start on startup
stop on shutdown
script
# We found $HOME is needed. Without it, we ran into problems
export HOME="/root"
exec sudo -u /usr/local/bin/node \
/var/www/vhosts/travelseguro.com/node/app.js \
2>&1 >> /var/log/node.log
end script
When I run sudo start app, I get:
start: Unknown job: app
How can I make this work?
回答1:
I was having the same problem running on the latest Amazon (AWS) linux which is Redhat based.
I have my upstart file in /etc/init called node.conf and when I ran sudo start node I would get a similar error to you start: Unknown job: node.
It turns out that the job won't start if there's an error in your .conf file. So I started out by commenting out all the lines and slowly building up to find the error. The error message isn't very clear and makes it look like upstart can't find your conf file.
Tailing your '/var/log/messages' will help you debug as Upstart logs to there (It may be somewhere different on Ubuntu. Mine said init: /etc/init/node-upstart.conf:8: Unknown stanza which helped me get to the bottom of it. In my particular case I was declaring variables incorrectly.
See on AskUbuntu for a similar thread.
Here's my edited working script:
<!-- language: lang-sh -->
#!upstart
# using upstart http://upstart.ubuntu.com/getting-started.html and node forever https://github.com/nodejitsu/forever/
# to run server
# redhat has additional sudo restrictions, you must comment out 'Default requiretty' from /etc/sudoers
#startup monitoring script derived from http://stackoverflow.com/questions/11084279/node-js-setup-for-easy-deployment-and-updating
description "node.js server"
author "jujhar"
env PROGRAM_NAME="node"
env FULL_PATH="/home/httpd/buto-middleman/public"
env FILE_NAME="forever.js"
env NODE_PATH="/usr/local/bin/node"
env USERNAME="springloops"
start on startup
stop on shutdown
script
export HOME="/root"
export NODE_ENV=staging #development/staging/production
echo $$ > /var/run/$PROGRAM_NAME.pid
cd $FULL_PATH
#exec sudo -u $USERNAME $NODE_PATH $FULL_PATH/$FILE_NAME >> /var/log/$PROGRAM_NAME.sys.log 2>&1
exec $NODE_PATH $FULL_PATH/$FILE_NAME >> /var/log/$PROGRAM_NAME.sys.log 2>&1
end script
pre-start script
# Date format same as (new Date()).toISOString() for consistency
echo "[`date -u +%Y-%m-%dT%T.%3NZ`] (sys) Starting" >> /var/log/$PROGRAM_NAME.sys.log
end script
pre-stop script
rm /var/run/$PROGRAM_NAME.pid
echo "[`date -u +%Y-%m-%dT%T.%3NZ`] (sys) Stopping" >> /var/log/$PROGRAM_NAME.sys.log
end script
-- Edit 2013-06-01 --
If you're on Centos or Amazon Linux like me, take a look at this init.d script.
-- Edit 2013-10-14 --
Here's a link to a gist of an init.d script that I actually use in production on Amazon Linux(Redhat Based). I simply keep it in my project under an init.d folder and then symlink to it in the /etc/init.d folder and now it's a daemon/service!
-- Edit 2014-06-05 --
Check out this awesome blog artcile by Jeff Dickey on Node.js in production using systemd which is much cleaner and easier than all the stuff we've been doing here (IMHO). He also uses Ansible to control his cluster (which I love) but you don't have to go that far if you're not ready.
回答2:
After a few attempts I implemented working .conf file for upstart which works as a service with automatic start after reboot and restart (respawn) in case of crash. Also it can start my app with unprivileged user permissions. The name of the file is /etc/init/my-app.conf.
To start / stop service please use
sudo start my-app / sudo stop my-app
If you have an error like
start: Unknown job: my-app
exec the following command
sudo initctl reload-configuration
My /etc/init/my-app.conf file:
#my-app
description "node.js my-app website"
env FULL_PATH="/home/myuser/app.prod/app"
env NODE_PATH="/usr/bin/node"
start on filesystem or runlevel [2345]
stop on [!2345]
script
export HOME="/root"
export NODE_ENV=production
echo $$ > /var/run/my-app.pid
cd $FULL_PATH
#Use exec below if you want to launch it under myuser,
#don't forget to create /var/log/my-app.sys.log with appropriate permissions
#exec sudo -u myuser sh -c "$NODE_PATH server.js >> /var/log/my-app.sys.log 2>&1"
exec $NODE_PATH server.js >> /var/log/my-app.sys.log 2>&1
end script
pre-start script
echo "[`date`] (sys) Starting" >> /var/log/my-app.sys.log
end script
pre-stop script
rm /var/run/my-app.pid
echo "[`date`] (sys) Stopping" >> /var/log/my-app.sys.log
end script
#uncomment respawn if you want to restart your service in case of crash
#respawn
#respawn limit 50 30
I do recommend to uncomment respawn after you will make sure that everything works ok.
UPDATE
I improved my script (please keep in mind that it works not under root but under regular user zn ):
#znapi.conf
description "node.js zn api"
env FULL_PATH="/home/zn/app.prod"
env NODE_PATH="/usr/bin/node"
env LOG="/var/log/znapi.log"
env PIDFILE="/var/run/znapi.pid"
# Start up when the system hits any normal runlevel, and
#start on filesystem or runlevel [2345]
#start when mongod started
start on started mongod
# shuts down when the system goes to shutdown or reboot.
stop on runlevel [06]
respawn
respawn limit 50 5
pre-start script
# check that log file exist or create it
test -f $LOG || { touch $LOG; chown zn:zn $LOG; }
# Date format same as (new Date()).toISOString() for consistency
echo "[`date`] (sys) Starting" >> $LOG
end script
script
export NODE_ENV=production
exec start-stop-daemon --start -m -p $PIDFILE -c zn -d $FULL_PATH -x server.js >> $LOG 2>&1
end script
pre-stop script
rm $PIDFILE
echo "[`date`] (sys) Stopping" >> $LOG
end script
来源：https://stackoverflow.com/questions/13982257/upstart-script-for-node-js-app