Hi Ken,
----- Original Message -----
> Changes committed to git://git.pcp.io/kenj/pcp master
>
> Ken McDonell (6):
> [...]
> src/pmlogger/rc_pmlogger: rework "chkconfig" logic
> [...]
> commit f24fe473f7de3a1eb859371c616a3d320d78287a
> Author: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
> Date: Tue Nov 17 08:05:10 2015 +1100
>
> src/pmlogger/rc_pmlogger: rework "chkconfig" logic
>
> If "chkconfig" (or the moral equivalent thereof, as hidden away
> in is_chkconfig_on() within rc-proc.sh) is "off", then do not
> start pmlogger. Previous behaviour was to warn, but start
> anyway.
>
> And when stopping, if "chkconfig" is "off" don't warn if a
> pmlogger process cannot be found, as that is now expected.
The current behaviour (warn but start anyway) is the result of many,
many people requesting this (rightly or wrongly) so we'll be getting
a slew of regression reports if this is changed yet again...
https://bugzilla.redhat.com/show_bug.cgi?id=1131022
https://bugzilla.redhat.com/show_bug.cgi?id=1185755
https://bugzilla.redhat.com/show_bug.cgi?id=1186012
(not to mention many direct requests not tracked anywhere).
The give-a-warning behaviour was the best compromise, I think - what was
the problem you've come across that led to this change? It will cause a
fair bit of angst if we go back to requiring chkconfig on, so if at all
possible I think we should stick with current warn-but-start behaviour.
This also affects pmie FWIW.
thanks.
--
Nathan
|