On 10/12/05, Saurabh Arora <[EMAIL PROTECTED]> wrote:
>
> Hi,
>
> I am using ganglia 3.0.1 and trying to monitor a cluster on unicast traffic
> with one node configured as a head node and other hosts reporting to it. I
> am using port 8651 and configured all nodes as "deaf" except the head node
> - the gmetad server pulls the data off from the head node. But I seem to get
> a broken display on the cluster output and some hosts seem to go down and
> come back up sporadically. Looks like either the server is not receiving the
> metrics from all the nodes (obviously via the hode node) or it isn't able to
> display it properly.
>
> The gmetad.conf/gmond.conf are attached. You can view the report at:
> http://dorado.umd.edu/ganglia
>
> gmond.conf (for the head node):
>
> /* This configuration is as close to 2.5.x default behavior as possible
> The values closely match ./gmond/metric.h definitions in 2.5.x */
> globals {
> setuid = yes
> user = nobody
> cleanup_threshold = 300 /*secs */
I believe somewhere it says in the list archives that 300 is a bad
value for this due to some internal setting. Try something different
-- I forget what values work.
> #deaf = true
> }
>
> /* If a cluster attribute is specified, then all gmond hosts are wrapped
> inside
>
> * of a <CLUSTER> tag. If you do not specify a cluster tag, then all
> <HOSTS> wi
> ll
> * NOT be wrapped inside of a <CLUSTER> tag. */
> cluster {
> name = "Milagro"
> }
>
> /* Feel free to specify as many udp_send_channels as you like. Gmond
> used to only support having a single channel */
> udp_send_channel {
> #mcast_join = 239.2.11.71
> host = ara.umd.edu
> port = 8651
> #mcast_if = eth0
> }
>
> /* You can specify as many udp_recv_channels as you like as well. */
> udp_recv_channel {
> #mcast_join = 239.2.11.71
> port = 8651
> #bind = 239.2.11.71
> #mcast_if = eth0
> }
>
> /* 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 = 8651
> }
>
> gmetad.conf
>
> data_source "Milagro" ara.umd.edu:8651
>
> #
> #-------------------------------------------------------------------------------
>
> # The name of this Grid. All the data sources above will be wrapped in a
> GRID
> # tag with this name.
> # default: Unspecified
> gridname "Milagro"
>
> # List of machines this gmetad will share XML with. Localhost
> # is always trusted.
> # default: There is no default value
> trusted_hosts 127.0.0.1 ara.umd.edu
>
> # If you want any host which connects to the gmetad XML to receive
> # data, then set this value to "on"
> # default: off
> all_trusted on
>
>
> Has anyone faced this before? I am running 3 more clusters and they seem to
> work just fine. Have a look at : http://cetus.umd.edu/ganglia and that works
Any differences between the clusters? (e.g. network topology, system
specifics, ganglia version, OS/distro, networking hardware, networking
method, etc?)
> perfect. But this one at http://dorado.umd.edu/ganglia is all screwed up.
>
> I'd appreciate if someone can help me out
>
> Thanks,
>
> Saurabh
>
>
>
--
~Mike
- Just my two cents
- No man is an island, and no man is unable.
