Today, Federico Sacerdoti wrote forth saying...

> So now that I am fresh on monday morning, I realize that gmetad is 
> statically linked to rrdtool, which is exactly right. So there is no 
> dependency between the rrdtool package and ganglia-monitor-core.

yep.  i was thinking of that as i walked into work today.
>
> I am writing a gmetad-webfrontend.spec file to build a webfrontend RPM, 
> which I will make dependent on the rrdtool package, since it calls 
> "rrdtool graph". Matt, do we need any particular version of rrdtool, or 
> will a generic dependence work?

if you build an RPM for the webfrontend you will also need to add a
dependency for apache and PHP too.  i don't think any particular version 
of rrdtool is necessary but then again i've only used two version so far.  
looking at the CHANGES file in 1.0.39, i think we'll be okay.  

> PS I wish we could call it "ganglia-webfrontend" since gmetad is not 
> very descriptive. But changing module names in CVS is a bitch.

i hate the name too.  i'm sure the name change can be done and i'm sure i 
don't know the right way to do it.  

the name "ganglia-webfrontend" works for me.  the only reason i named it 
the webfrontend was because i can forsee many ganglia "frontends".  there 
is one group in japan who email'd me that they are writing a GTK frontend 
(whether it really happens or not we'll see).  another possibility is to 
call it the "ganglia-portal" since i think portals are uniquely web-based.

-matt


Reply via email to