Re: [Fancontrol] better support for init script error handling

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi,

On Sat, 01 May 2010 17:29:04 +0200, PyroPeter wrote:
> Some months ago, the /etc/fancontrol syntax changed,

Please be specific. The syntax never changed, it was extended in
fully backwards compatible ways, this should never have caused
fancontrol to fail, unless you downgraded it.

> and fancontrol 
> exited right after startup, but the init script stated success.
> 
> Currently there is no reliable way for a init script to check if 
> fancontrol started up properly.

Running it using startproc should do the trick. That's what openSUSE
does.

> As fancontrol is a deamon preventing hardware damage, reliability should 
> be of first priority.

This is incorrect. fancontrol is a daemon _causing_ hardware damage, if
anything. The initial state of your system should be safe, so
fancontrol not starting should never be a safety issue.

> I would suggest adding a argument to fancontrol that makes it start up, 
> check config syntax and write permissions, and than fork to the 
> background, or return a positive exit code.
> 
> 
> I wrote a kind of proof of concept that simply uses the existing 
> bash-script and "forks" by reexecuting itself in the background:
> 
>  >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>  >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
> 
> diff -ru lm_sensors-3.1.2-1/usr/sbin/fancontrol 
> lm_sensors-3.1.2-1_pyropeter/usr/sbin/fancontrol
> --- lm_sensors-3.1.2-1/usr/sbin/fancontrol    2010-02-03 
> 03:45:15.000000000 +0100
> +++ lm_sensors-3.1.2-1_pyropeter/usr/sbin/fancontrol    2010-03-07 
> 01:37:09.000000000 +0100
> @@ -5,7 +5,9 @@
>   #
>   # Version 0.70
>   #
> -# Usage: fancontrol [CONFIGFILE]
> +# Usage: fancontrol [-D] [CONFIGFILE]
> +#
> +# (-D causes fancontrol to 'fork' to the background after some tests)
>   #
>   # Dependencies:
>   #   bash, egrep, sed, cut, sleep, readlink, lm_sensors :)
> @@ -43,6 +45,12 @@
>   #DEBUG=1
>   MAX=255
> 
> +DAEMON=0
> +if [ "$1" = "-D" ]; then
> +    DAEMON=1
> +    shift
> +fi
> +
>   declare -i pwmval
> 
>   function LoadConfig {
> @@ -303,7 +311,6 @@
>       echo "File $PIDFILE exists, is fancontrol already running?"
>       exit 1
>   fi
> -echo $$ > "$PIDFILE"
> 
>   # $1 = pwm file name
>   function pwmdisable()
> @@ -475,6 +482,14 @@
>       let fcvcount=$fcvcount+1
>   done
> 
> +if [ "$DAEMON" -gt 0 ]; then
> +    echo "Forking..."
> +    $0 $* &> /dev/null &
> +    exit 0
> +fi
> +
> +echo $$ > "$PIDFILE"
> +
>   echo 'Starting automatic fan control...'
> 
>   # main loop calling the main function at specified intervals
> 
>  >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>  >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
> 
> I am not subscribed to this mailing list, so you need to CC me.
> 

This means that the checks will be done twice. And if the checks are
somehow incomplete, fancontrol may still fail despite the fork being
successful, so it is unreliable by design.

If startproc works for you, you should use it.

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

[Index of Archives]     [Linux Kernel]     [Linux Hardware Monitoring]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux