upstart, exec and stderr

Question

I am using syslog-ng on a Ubuntu Lucid machine with the following upstart script:

# syslog-ng - system logging daemon
#

description     "Syslog-ng daemon"

start on (local-filesystems and net-device-up IFACE!=lo)
stop on runlevel [!2345]

expect fork
respawn

pre-start script
    test -x /usr/sbin/syslog-ng || { stop; exit 0; }
    mkdir -p -m0755 /var/run/syslog-ng
end script

exec /usr/sbin/syslog-ng -p /var/run/syslog-ng/syslog-ng.pid

When the syslog-ng configuration file is wrong, syslog-ng outputs an error message on stderr. Unfortunately, this error is caught by upstart and doesn't get to the console when starting the service, so there is no way to know why the service start failed.

Is it normal that upstart would catch stderr? Can it be set?

Answer 1

This is documented in init(5); you can add the following line to your /etc/init/syslog-ng.conf to see output:

console output

---

Alternatively, you can see output from all upstart scripts when you add the following to the kernel boot parameters:

INIT_VERBOSE=yes

You can do that temporarily by editing the parameters in the grub menu during boot, or (more) permanently by editing /etc/default/grub and adding it to GRUB_CMDLINE_LINUX, then afterwards run update-grub.

For more information and best practices, see The Upstart Cookbook.

Answer 2

If the stdout is getting through what you can do is. which will pipe the stderr to stdout which should solve your problem. :)

syslog-ng 2>&1