Hi Ondřej,
On Fri, May 31, 2013 at 11:44:53PM +0200, Ondřej Surý wrote:
> I have tried to rewrite php5-fpm init.d file for systemd and upstart and
> ended up with:
> cat > php5-fpm.service << EOF
> [Unit]
> Description=The PHP FastCGI Process Manager
> After=syslog.target network.target
> [Service]
> Type=forking
> PIDFile=/var/run/php5-fpm.pid
> ExecStartPre=sh -c 'if [ -n "$(/usr/sbin/php5-fpm --fpm-config
> /etc/php5/fpm/php-fpm.conf -t 2>&1 | grep "\[ERROR\]")" ]; then echo
> "Please fix your configuration file..."; /usr/sbin/php5-fpm --f\
> pm-config /etc/php5/fpm/php-fpm.conf -t 2>&1 | grep "\[ERROR\]"; exit 1; fi'
> ExecStart=/usr/sbin/php5-fpm
> ExecReload=/bin/kill -USR2 $MAINPID
> [Install]
> WantedBy=multi-user.target
> EOF
> and
> cat > php5-fpm.upstart << EOF
> # php5-fpm - The PHP FastCGI Process Manager
> description "The PHP FastCGI Process Manager"
> author "Ondřej Surý <ondrej@debian.org>"
> start on (local-filesystems and net-device-up IFACE!=lo)
> stop on runlevel [016]
FYI, it's strongly recommended to use 'start on runlevel [2345]' here as the
start condition, for several reasons:
- The 'filesystem' events are one-time events seen only at boot time; if
the admin drops the system to runlevel 1 and then returns to runlevel 2,
you probably want the service to restart.
- The event name is 'local-filesystem', not 'local-filesystems', so this
job wouldn't auto-start at all. (FWIW, the 'initctl check-config'
command from the upstart package would warn about this; but this tool is
only usable when upstart is the running init, so probably not ideal for
packagers not running upstart themselves.)
- The 'local-filesystem' event may be emitted before any remote filesystems
have been mounted... which might include part or all of /usr and /var.
So since php5-fpm is itself not a service that is used for mounting
remote filesystems, it should not try to start until the filesystem is
completely up.
The 'runlevel' event is later than both the 'local-filesystem' event and the
'net-device-up' events, except in the case where your network interface is
configured with network-manager instead of ifupdown. So you shouldn't need
to worry about this causing your service to start too "early".
> pre-start script
> if [ -n "$(/usr/sbin/php5-fpm --fpm-config /etc/php5/fpm/php-fpm.conf -t 2>&1 | grep "\[ERROR\]")" ]; then
> echo "Please fix your configuration file..."
> /usr/sbin/php5-fpm --fpm-config /etc/php5/fpm/php-fpm.conf -t 2>&1 |
> grep "\[ERROR\]"
> stop ; exit 1
> fi
> end script
Unrelated style comment - you could write this to call php5-fpm -t a single
time:
pre-start script
errors=$(/usr/sbin/php5-fpm --fpm-config /etc/php5/fpm/php-fpm.conf -t 2>&1 | grep "\[ERROR\]")
if [ -n "$errors" ]; then
echo "Please fix your configuration file..."
echo $errors
stop ; exit 1
fi
end script
Also, the 'exit 1' is unnecessary; the 'stop' is sufficient to ensure the
job stops, all the 'exit 1' does is result in additional log spam.
> respawn
> exec /usr/sbin/php5-fpm --fpm-config /etc/php5/fpm/php-fpm.conf
> EOF
> It's pretty much equivalent with one exception – I need to send USR2 on
> reload. Does upstart already have the support for custom reload signals?
Presently not, as Clint mentions - though I see Dmitrijs has just claimed
https://bugs.launchpad.net/upstart/+bug/893021 for himself in response to
this thread. :)
--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
slangasek@ubuntu.com vorlon@debian.org
Attachment:
signature.asc
Description: Digital signature