Bernard, would this be in addition to running 'gmetad' on the head node
('redman') or replacing 'gmetad' on the head node?
What I'm seeing right now - with 'gmond' running on all nodes and the head
node, using the identical 'gmond.conf' file (except that on the head node,
'mute = yes') - is that all of my nodes can see each other, but the head
node only sees itself... when I "telnet localhost 8654".
The head node has two interfaces - public and private - and the nodes are
all on the private interface.
Thanks.
Best,
Stephen
On Thu, Mar 25, 2010 at 12:16 PM, Bernard Li <[email protected]> wrote:
> Hi Stephen:
>
> On Thu, Mar 25, 2010 at 10:59 AM, Stephen Spencer
> <[email protected]> wrote:
>
> > I have an existing Ganglia installation with two data_sources,
> corresponding
> > to two homogeneous sets of cluster nodes.
> > The relevant lines from /etc/ganglia/gmetad.conf on the head node of this
> > cluster - call it 'frame' - are:
> >
> > data_source "no-gpu" node.fqdn (using the default port)
> > data_source "gpu" node.fqdn:8653
> >
> > I'm deploying a new cluster - call it 'redman' - and would like to have
> > THESE nodes show up as a third data source. I'll have 'gmond' on each of
> the
> > nodes, and 'gmetad' on the head node, just like I do on the original
> > cluster.
> >
> > How should I set up data collection on 'redman,' and how can I configure
> > 'frame' to retrieve that collected data from 'redman'? My first thought
> > would be to run 'gmond' on 'redman'... but I don't want data from
> 'redman'
> > to be collected; it's a head node, not a compute node.
> >
> > The head nodes of each cluster - 'frame' and 'redman' - are on the public
> > network, but their nodes are not.
>
> Run gmond on 'redman', but set "mute" to "yes", then it should not
> collect any data. See the gmond.conf manpage for additional
> information.
>
> Cheers,
>
> Bernard
>
--
Stephen Spencer
[email protected]
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general
