Red Hat Bugzilla – Bug 1131022
pmlogger service not starting when not chkconfig'ed on
Last modified: 2015-07-22 02:24:27 EDT
Description of problem: service pmlogger start exits quietly unless pmlogger hasn't been chkconfig'ed on, it should be possible to start pmlogger without enabling it permanently. # service pmlogger status Checking for pmlogger: stopped # service pmlogger start # sleep 10 # service pmlogger status Checking for pmlogger: stopped # chkconfig pmlogger on # service pmlogger start Starting pmlogger ... # sleep 10 # service pmlogger status Checking for pmlogger: running #
I agree we should definitely allow this - I think the current behaviour is a remnant from other operating systems (IRIX). There's code there to issue a diagnostic but its either not being activated (pmie case) or not all there (in the pmlogger case) - both should be fixed up. Tagging this for 6.7 and will fix upstream soon - thanks Marko.
There's some concern / on-going discussion upstream about the consequences of starting pmloggers that are not being properly tracked (in terms of not having logs rotated, compressed, and so on) and things like cronjobs and helper scripts being closely reliant on chkconfig state to know whether they should operate. I'll keep on working on it (we have a separate bug open as a systemd-variant of this BZ), but there's no short-term easy solution. Since some deadlines are close for RHEL6.7 and I think we should definitely do something to help the major problem here (i.e., the service silently failing to start) - so I've tackled the diagnostics/reporting side of the problem in the interim, as follows: # /sbin/chkconfig pmlogger off # service pmlogger start /etc/rc.d/init.d/pmlogger: Warning: Performance Co-Pilot archive logger(s) are disabled. To enable pmlogger, run the following as root: # /sbin/chkconfig pmlogger on # /sbin/chkconfig pmlogger on # service pmlogger start Starting pmlogger ... # So, the punter no longer has to spend ages trying to figure out why the start script didn't do anything. I'll track the remainder of the problem under the Fedora BZ, and a future rebase for RHEL will pick up other work done there to further improve the situation, as soon as it becomes available.
Verified for build pcp-3.10.3-3.el6.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2015-1300.html