[Pacemaker] More logging Issues

[imnotpc]

On Sunday, June 26, 2011 19:16:58 Andrew Beekhof wrote:
> On Thu, Jun 23, 2011 at 2:37 AM, imnotpc <imnotpc at rock3d.net> wrote:
> > On Tuesday, June 21, 2011 23:15:15 Andrew Beekhof wrote:
> >> On Wed, Jun 22, 2011 at 9:16 AM, imnotpc <imnotpc at rock3d.net> wrote:
> >> > As I'm grinding my way through the configuration of my first cluster,
> >> > I've resolved my notification issues with crm_mon and a custom script.
> >> > Now I'm trying to get logging set up and I have a couple more
> >> > questions.
> >> > 
> >> > After reading the list archives, search results, and the docs that
> >> > I've found (or have been pointed out to me, thanks), it seems that
> >> > pacemaker doesn't have any log configuration options and always logs
> >> > to syslog. Is this right or am I missing something?
> >> 
> >> It reuses the logging options set in cluster.conf, corosync.conf or
> >> ha.cf depending on how pacemaker was started
> > 
> > Yes it follows the global logging configuration in cluster.conf.
> > 
> >> > Following the cluster.conf man page I set up cman/corosync logging and
> >> > it worked as expected. However after running pacemaker on these nodes
> >> > logging seemed to revert to original settings. I thought it might be
> >> > a cache issue so I rebooted but it still ignores the new settings. Is
> >> > this a coincidence or can pacemaker actually override manual changes
> >> > to cluster.conf?
> > 
> > This is still not working though. I tried re-installing everything and
> > now corosync and the pacemaker daemons follow the logging configuration
> > in cluster.conf. dlm_controld, fenced, and gfs_controld follow the
> > global syslog settings but ignore the global to_logfile="no" setting and
> > continue to log to their default log files. Here's the only logging line
> > in my cluster.conf which validates with ccs_config_validate:
> > 
> > <logging syslog_priority="debug" to_syslog="yes" to_logfile="no"/>
> > 
> > I've also tried using the logging_daemon setting to configure each daemon
> > individually but this setting is also ignored.
> 
> dlm_controld, fenced, and gfs_controld are not part of pacemaker - i
> don't know where they get their logging setup from or even if syslog
> is supported.

Ok, thanks. I'm still trying to grasp what daemon belongs to what package and
who maintains it. It seems those daemons follow the global syslog settings in
cluster.conf but don't parse the logfile settings correctly. I've resolved the
logging issue by ignoring the default log files and writing everything to 
syslog at the debug level then parsing the input and redirecting to the output 
to my log files at the desired logging level.

Jeff