Bernard,
On Mar 4, 2011, at 3:30 PM, Bernard Li wrote:
> Hi Neil:
>
> On Fri, Mar 4, 2011 at 1:35 PM, Neil McKee <neil.mc...@inmon.com> wrote:
>
>> OK, I added some commented-out lines to lib/default_conf.h.in, and checked
>> in the changes in that directory. Should be OK now(?)
>
> Looks fine.
>
> BTW, did we already decide that 'accept_all_physical' is no longer a
> configurable option?
>
I wasn't sure about this. Do you think anyone would ever set
"accept_all_physical = no"? And what would this option mean if the base set
in libmetrics was extended? It seems like a moving target. Hard to describe.
From what I can tell the UI does a great job of accepting extra metrics, so I
left the option out for now. Should I put it back?
> The configuration option in gmond.conf seems to be increasing over
> time -- I wonder if it would be better to put the sflow configuration
> in a separate file, such as /etc/ganglia/conf.d/sflow.conf. It will
> get included via the "include" directive in gmond.conf. What do you
> think?
Not sure if that is really necessary at this point. In most cases there won't
be any need for an sflow { } section at all. The only time you would use it is
to listen on a non-standard port (rare) or to allow hypervisors to report on
behalf of their VMs (also rare, I suspect).
>
> I noticed something weird with the metrics from my WinXP host (running
> hsflowd):
>
> The gmond collector was down for say an hour, during that time,
> hsflowd on WinXP was still running. However, when I started gmond up,
> I noticed that the "boottime" metric was no longer a constant, and
> changes with time. As a result, "Uptime" and "Last Reported" are
> quite close to each other. I also noticed that certain metrics like
> load_* were not correct either (lower than what it should be).
>
> I haven't really been able to re-produce this, but thought I'd ask you
> to see if you know why that would happen. Will definitely write back
> if I encounter it again in the future.
Definitely a head-scratcher, that one. I'll try to reproduce it too.
Neil
>
> Cheers,
>
> Bernard
------------------------------------------------------------------------------
What You Don't Know About Data Connectivity CAN Hurt You
This paper provides an overview of data connectivity, details
its effect on application quality, and explores various alternative
solutions. http://p.sf.net/sfu/progress-d2d
_______________________________________________
Ganglia-developers mailing list
Ganglia-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ganglia-developers
