Database Reference
In-Depth Information
Ganglia
Like the HBase metric framework, it also inherits direct support for Ganglia
from Hadoop.
Ganglia (
http://ganglia.sourceforge.net/
) is a scalable
distributed, monitoring system for high-performance computing
systems, such as clusters and grids. It was developed at the
University of California Berkeley and open sourced. The Hadoop
and HBase communities have been using it as the de facto solution
to monitor clusters.
Ganglia has three important components, which are as follows:
• The Ganglia monitoring daemon (
gmond
): This daemon needs to run on every
machine that is monitored. It monitors the host and collects the local data
and prepares the statistics to be polled by other systems. The
gmond.conf
ile deines the properties required for the
gmond
daemon as follows:
globals {
user = ganglia
}
cluster {
name = "hbase-Test"
}
udp_send_channel {
host = localhost
port = 8649
}
udp_recv_channel {
port = 8649
}
• The Ganglia meta daemon (
gmetad
): This daemon is installed on a central
node and acts as the federation node to the entire cluster. It polls monitoring
data from running
gmond
daemons within the cluster and saves it as a
time-series database. This data can further be used as XML by its own
web client or any other third-party client. The
gmetad.conf
ile deines
the properties required for the
gmond
daemon as follows:
data_source "hbase-Test" localhost
gridname "hbase-Test"
setuid_username "ganglia"
