Problem solved.
As it turns out, it was determined that the device was dropping UDP packets (as
shown by sysctl).
The solution was to increase the net.core.rmem_max and gmond (via gmond.conf
udp_recv_channel's buffer).
Btw: I upgraded Ganglia to 3.6.0 which supports the buffer option for gmond
Does anyone know where to remove some of the default metrics from the gweb
metrics pull-down menu?
Metrics such as apache_report and varnish_report?
--
Slashdot TV. Videos for Nerds. Stuff that Matters.
the parameters and AFAIK, there is nothing in Ganglia 3.x
that controls or limits metrics.
I've also tried increasing server_threads in gmetad.conf.
From: Dan (Daniel) Arsenault [mailto:dar...@broadcom.com]
Sent: Tuesday, September 23, 2014 2:00 PM
To: ganglia-developers@lists.sourceforge.net
Subject: Re
I am using Ganglia 3.6.
I am using Python to create custom metrics.
I am using a loop to generate the metrics with unique names.
I appears that there is a limit that is preventing all the metrics to be shown
on gweb.
There is no problem showing a smaller number of metrics (around 50 or so).
But
All the metrics are within the same 'group' and when they reach a certain count
and are not displayed on the web.
I checked the gmond messages using netcat as suggested and the missing metrics
are present.
From: Dan (Daniel) Arsenault [mailto:dar...@broadcom.com]
Sent: Tuesday, September 23