[collectd] Collectd redundancy infrastructure

[Cyril Feraudet]

Hi all, 

I've more than 3000 servers on 15 datacenters monitored successfully with
collectd. 

All data center have now 2 proxys who relay datas to the main server : 

Servers :  
<Plugin network>
	Server "local-proxy-1" "25826"
	Server "local-proxy-2" "25826"
</Plugin>

Proxys : 
<Plugin network>
	Listen "local-proxy-x" "25826"
	Server "main-server" "25826"
	Forward true
	ReportStats true
</Plugin>

Main servers :
<Plugin network>
	Listen "main-server" "25826"
	ReportStats true
</Plugin>

In this case all datas are received twice for each server and collectd
(through rrdcached) try to write twice on disk and report an rrd error like
: illegal attempt to update using time 1312895973 when last update time is
1312895973 (minimum one second step) / rrdcached plugin: rrdc_update
(/var/lib/c...., [1312896089:0.000000:0.000000:0.560000], 1) failed with
status -1.

How to make collectd ignore datas already sent ?

Regards,

Cyril