All, I'm using the write_graphite plugin to write to a graphite cluster. The data I'm concerned with is coming from a central "polling" system that, via SNMP or the Netapp plugin, queries other appliances incapable of running the collectd client. I need this data to reside in a different directory when written based on certain host attributes. On a regular system, I simply add a "Prefix" to write_graphite and that puts the data in the correct place. On a polling machine, that forces the data for the polled system to have the same prefix added, putting it in the wrong directory.
It appears that this should be resolvable via chains, but I'm not having any success. In the write_graphite documentation, the example configuration is described as registering "a writer under this string: write_graphite/example" because of the <Node "example"> ... </Node> configuration. To that end, I've created a write_graphite configuration with multiple nodes and different prefixes: <Plugin write_graphite> <Node "default"> Host "graphite-vip" Port "2003" Prefix "core.wl.syseng-test.dev." SeparateInstances true </Node> <Node "netapp"> Host "graphite-vip" Port "2003" Prefix "core.sh5.storage-netapp.prd." SeparateInstances true </Node> <Node "netscaler"> Host "graphite-vip" Port "2003" Prefix "core.na.netscaler-lb." SeparateInstances true </Node> </Plugin> How can I access these different nodes as separate writers? My chain configuration is complaining that they're invalid following the syntax in the wiki: LoadPlugin "match_regex" LoadPlugin "target_write" <Chain "PreCache"> <Rule "rewrite_netapp"> <Match "regex"> Plugin "netapp" </Match> <Target "write"> Plugin "write_graphite/netapp" </Target> Target Stop </Rule> <Rule "rewrite_netscaler"> <Match "regex"> Plugin "snmp" </Match> <Match "regex"> Host "^netscaler.*" </Match> <Target "write"> Plugin "write_graphite/netscaler" </Target> Target Stop </Rule> # Default target <Target "write"> Plugin "write_graphite/default" </Target> </Chain> Question: Is this possible? And if so, what am I doing wrong? Thanks, Mark
_______________________________________________ collectd mailing list collectd@verplant.org http://mailman.verplant.org/listinfo/collectd