For one of our small 3 node clusters it is setup something like this (the guy
who set it up is on vacation):
Node 1:
udp_send_channel {
#bind_hostname = yes # Highly recommended, soon to be default.
# This option tells gmond to use a source address
# that resolves to the machine's hostname. Without
# this, the metrics may appear to come from any
# interface and the DNS names associated with
# those IPs will be used to create the RRDs.
host = ip of head node
port = 8666
}
/* You can specify as many udp_recv_channels as you like as well. */
udp_recv_channel {
port = 8666
}
/* You can specify as many tcp_accept_channels as you like to share
an xml description of the state of the cluster */
tcp_accept_channel {
port = 8649
}
----
Node 2:
/* Feel free to specify as many udp_send_channels as you like. Gmond
used to only support having a single channel */
udp_send_channel {
#bind_hostname = yes # Highly recommended, soon to be default.
# This option tells gmond to use a source address
# that resolves to the machine's hostname. Without
# this, the metrics may appear to come from any
# interface and the DNS names associated with
# those IPs will be used to create the RRDs.
host = ip of head node
port = 8666
}
/* You can specify as many udp_recv_channels as you like as well. */
udp_recv_channel {
port = 8666
}
/* You can specify as many tcp_accept_channels as you like to share
an xml description of the state of the cluster */
tcp_accept_channel {
port = 8649
}
----
Head Node:
/* Feel free to specify as many udp_send_channels as you like. Gmond
used to only support having a single channel */
udp_send_channel {
#bind_hostname = yes # Highly recommended, soon to be default.
# This option tells gmond to use a source address
# that resolves to the machine's hostname. Without
# this, the metrics may appear to come from any
# interface and the DNS names associated with
# those IPs will be used to create the RRDs.
host = 127.0.0.1
port = 8666
}
/* You can specify as many udp_recv_channels as you like as well. */
udp_recv_channel {
port = 8666
}
/* You can specify as many tcp_accept_channels as you like to share
an xml description of the state of the cluster */
tcp_accept_channel {
port = 8649
}
And in the gmetad:
data_source "cluster name" localhost
----
I'm guessing with this setup all the other gmonds in the cluster send their
data via UDP unicast to the gmond located on the head node which acts like a
hub, which connects via TCP to the gmetad located on the same machine and
relays the data?
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general
