I don't have the hardware to test. Try this and report if it works.
Replace:

[ -z "$PID" ] && /usr/sbin/fancontrol >/dev/null 2>&1 &
if [ $? -gt 0 ]; then

by:

[ -z "$PID" ] && /usr/sbin/fancontrol >/dev/null 2>&1 &
if [ $? -gt 0 -o -z "$PID" ]; then

No, it didn't work. With that it reports fail even if it starts successfully.

I'll probably need to ask another dev to look into it. Just curious: If you change it back to what it was before the previous bug report, does it work?

[ -z "$PID" ] && /usr/sbin/fancontrol >/dev/null 2>&1 &
if [ -z "$PID" -o $? -gt 0 ]; then

With that it says it fails even though it starts successfully.

This works for me:

-------
if [ $? -gt 0 -o -n "$PID" ] ; then
-------

But it does not output a "fail" if fancontrol actually fails. (This is caused by starting the deamon in the background, which makes it impossible to determine its exit-status)

The only good way to fix this is to modify the fancontrol script.
It should get a "-d" argument, which makes it start, run all checks, then crash and return >0, or fork to the background. (I don't have a clue if bash is capable of forking in a script, but as a workaround it could re-call itself in the background and exit. the second instance should most likely find valid working conditions.)

Here is a patch that adds the above mentioned commandline argument to fancontrol.

Its just a kind of POC and not tested very well.

My current initscript using the patched fancontrol looks like this:

------
PID=$(pidof -o %PPID -x /usr/sbin/fancontrol2)
case "$1" in
start)
stat_busy "Starting fancontrol"
[ -z "$PID" ] && /usr/sbin/fancontrol2 -D &>/dev/null
if [ $? -gt 0 -o -n "$PID" ] ; then
------

Those with the problem-- what sort of hardware _do_ we need to test this? I admit to not knowing much about lm_sensors, but if I have relevant hardware, I can try to help.

I think lm_sensors is mostly used for CPUs or mainboards. If you use a CPU in your computer, you should take a look at http://www.lm-sensors.org/wiki/Devices .

Attached is an improvment of the above posted patch. (It also covers the initscript)