Dave Bradshaw wrote:
Dear All,
Am I an idiot?
My rule of thumb is not to ask this sort of question on a list unless
I'm asking something already covered in the docs. There's always a
chance some wisecracker out there will answer it.
:)
> Where am I going wrong?
Different clusters need to transmit on different multicast IPs or ports.
Otherwise, they will all hear each other's metrics.
So cluster A can use 232.4.6.11:8649, cluster B can use 232.4.6.12:8649,
cluster C can use 232.4.6.12:8648 ... none of them should overlap, in
that case.
The cluster name has no real effect on the way the monitoring core
gathers metrics. The metadaemon may use it (in comparison to the source
name in /etc/gmetad.conf), but I don't remember what the logic there
looks like.
All the metadaemon does is query a monitoring core per data_source line
in the config, parse the results, store them as RRDs and merge all its
data sources into a single XML document for the web front-end to query.
Okay, so it does a lot, but my point is that the problem you're having
is upstream.
Only one host (or, in very large installs, a couple hosts) need to run
gmetad. The others just need to run the monitoring core.
Good luck!
