Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have following upstart job:

description "posty api"

start on mysql
stop on shutdown

env RACK_ENV=production

setuid vmail
setgid vmail

chdir /opt/posty_api

pre-start script
    mkdir -p /var/run/posty
    chown -R vmail:root /var/run/posty
end script

exec /usr/local/bin/unicorn -D -c /opt/posty_api/unicorn.rb --env production >> /var/log/posty/upstart.log 2>&1

post-stop exec kill `cat /var/run/posty/`

respawn limit 1 10

To create the folder in /var/run I need root privileges. How can I run some parts of the upstart job as root, and the service itself as unprivileged user?

share|improve this question
up vote 4 down vote accepted

upstart has no facility equivalent to systemd's PermissionsStartOnly setting. All processes in the job run as the user set via the setuid stanza, as the Cookbook says.

So do things the daemontools way.

Use setuidgid, setuidgid, s6-setuidgid, chpst, runuid, or setuidgid in the exec stanza:

exec \
setuidgid somebody \
unicorn -D -c /opt/posty_api/unicorn.rb --env production >> /var/log/posty/upstart.log 2>&1

That's a terrible logging mechanism, by the way. The daemontools way would have a proper, automatically cycled, rotateable-on-demand, size-capped, log using multilog, multilog, s6-log, svlogd, tinylog, or cyclog. upstart is tricky to integrate with those, however, given its expect mechanism.

expect fork
exec \
setuidgid somebody \
unicorn -D -c /opt/posty_api/unicorn.rb --env production 2>&1 | \
/usr/local/bin/chdir /var/log/ \
setuidgid log \
cyclog posty/unicorn/

(The chdir here is the chain-loading one from the nosh package, and isn't strictly necessary. But it makes things somewhat tidier.)

share|improve this answer
Thx for the explanation and the hint for logging. I created already two upstartjobs, the prepare job referencing with start on starting posty. But the logging is way better than mine. – Waxolunist Mar 2 '14 at 14:08

you can use setguid in script block and it will affect only given block. Something like this:

# Ubuntu upstart file at /etc/init/kafka.conf
description "kafka broker"

limit nofile 32768 32768

start on runlevel [2345]
stop on [!2345]

respawn limit 2 5

umask 007

kill timeout 300

pre-start script
    # here you are root
    ls -alh /root/
end script

chdir /usr/local/lib/kafka

    setuid kafka
    setgid kafka
    # here you are user/group kafka
    /usr/local/lib/kafka/bin/ /usr/local/lib/kafka/config/
end script
share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.