Here is the bug with the web-frontend patch for 3.1.7
http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=279
This patch helped me to pick and choose which custom MySQL graphs to display
for each host, an entire cluster, and even the whole grid. The code has
comments all over the place. If you don't understand any part of it, let me
know.
Note: The second attachment includes the conf.php config. You want that one.
--
Jamie Isaacs
[email protected]
On Fri, Oct 8, 2010 at 4:01 PM, CHU, STEPHEN H (ATTSI) <[email protected]>wrote:
> Hi Bernard,
>
> Collectl provides two alternatives to make this happen. See
> http://collectl.sourceforge.net/Gexpr.html. I chose the method of
> supplementing the gmond data with the collectl data. Fairly straight
> forward. You can also run collectl alone on a server and feed the data
> to a centralized gmond collector and back to gmetad.
>
> Steve
>
> -----Original Message-----
> From: Bernard Li [mailto:[email protected]]
> Sent: Friday, October 08, 2010 4:02 PM
> To: CHU, STEPHEN H (ATTSI)
> Cc: [email protected]
> Subject: Re: [Ganglia-general] Custom graphs only for specific Cluster
>
> Hi Steve:
>
> On Fri, Oct 8, 2010 at 12:03 PM, CHU, STEPHEN H (ATTSI) <[email protected]>
> wrote:
>
> > I have set up 3 different clusters all homed in to a gmetad server and
> > served by a webfrontend. One of the cluster is a Lustre file system
> with 6
> > servers. To gather Lustre specific data I use collectl which supports
> > exporting Lustre data to Ganglia. Collectl is running on each of the
> Lustre
> > server alongside with gmond. Collectl feeds exported XML data to gmond
> UDP
> > port and the rrd data get delivered back to the gmetad server. I like
> to
> > display these data but they don't belong to any metrics defined. After
> > reading up on how to generate custom graphs, I have successfully
> created
> > graphs for them (keying on some existing metrics like
> bytes_in/bytes_out)
> > and they are displayed under the Lustre cluster page. However, when I
> select
> > the other clusters, they are also being "shown" with no graphs but
> with the
> > name of the custom graph report. That's understandable since they
> don't have
> > any of the Lustre data.
>
> Wouldn't it be more efficient if you simply port the Lustre metric
> collection code to gmond via C/Python modules and/or run collectl only
> on the hosts and send all the data to a collector gmond? That way you
> don't need to run both gmond and collectl on your hosts.
>
> Just my $0.02.
>
> Cheers,
>
> Bernard
>
>
> ------------------------------------------------------------------------------
> Beautiful is writing same markup. Internet Explorer 9 supports
> standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3.
> Spend less time writing and rewriting code and more time creating great
> experiences on the web. Be a part of the beta today.
> http://p.sf.net/sfu/beautyoftheweb
> _______________________________________________
> Ganglia-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/ganglia-general
>
------------------------------------------------------------------------------
Beautiful is writing same markup. Internet Explorer 9 supports
standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3.
Spend less time writing and rewriting code and more time creating great
experiences on the web. Be a part of the beta today.
http://p.sf.net/sfu/beautyoftheweb
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general
