Chris Stone wrote:
Ganglia is great. I got it up and running on my linux cluster in short
order. I do have one nagging detail I'd like to remedy.
/var/lib/ganglia/rrds/ contains a directory called "unspecified". My
ganglia web page also lists this name as the name of the cluster, ie.
"unspecified> node1". All of my node names show up correctly, and all of
the details seem correct. I'd like to change the "unspecified" to
something specific, but I can't find where to do that. I'm assuming the
change should be made in /etc/gmetad.conf. Can some one point me in the
right direction.
Close, but no cancer stick.
[disclaimer: My answers are occasionally wrong.]
Check the configuration of the monitoring core on the machine(s) acting as
your data source. gmetad overrides the label specified in its
configuration with whatever's in the XML stream that it collects from the
monitoring core.
For example, if you have a data source defined with the label "rabbits" and
you're polling hosts 'binky,' 'bongo,' and 'sheba' on it, and 'binky'
thinks the cluster is named 'WorkIsHell' and 'bongo' thinks the cluster is
named 'SchoolIsHell' and 'sheba' thinks the cluster is named 'LoveIsHell',
gmetad will label the cluster (and the RRD directory, confusingly enough)
as 'WorkIsHell' if it polls 'binky,' 'SchoolIsHell' if it polls 'bongo,'
and so forth. To fix it, you would have to change the cluster name in
/etc/gmond.conf and restart the monitoring daemon on each host you're
actively polling with gmetad.
Of course, if you only poll one host, you only have to change it once...
Something tells me my simple explanation actually makes this more
complicated -- oh well. :)
That reminds me, Matt & co., have we thought about trapping SIGHUP so's we
can do a config re-read? getconf supports the concept, doesn't it? Might
be nice to find a way to avoid repopulating the metric hash from scratch
every time (maybe only if certain config values change?) ...
Or maybe something snazzier for Ganglia 3...
