search

nytimes / collectd-rabbitmq

A collected plugin, written in python, to collect statistics from RabbitMQ.
https://collectd-rabbitmq.readthedocs.org/
Other
145 stars 79 forks source link

Format issue with graphite output #42

Open smetj opened 8 years ago

smetj commented 8 years ago

Reopening issue https://github.com/NYTimes/collectd-rabbitmq/issues/8

When outputting via write_graphite plugin, the node stats are outputted correctly, e.g.: server1.rabbitmq.disk_free

However, the exchange and queue stats do not prefix the server name correctly: rabbitmq_default.exchanges-logstash_backfill1.ack

Not sure why the servername would show up on one but not the other?

Still ticket was closed with the remark:

queue/exchange stats are shared across nodes in a cluster, rather than reported per node.

prefixing with hostname or not, at least there must be an option to prefix the rabbitmq_default namespace (without using the collectd global namespace) because if you are collecting from multiple clusters (or multiple individual nodes) and writing the metrics in the same (graphite) backend then you end up unintentionally merging metrics into 1 namespace.

redterror commented 8 years ago

Suggestion - use the cluster_name returned in the overview stats here? Since these are cluster-wide, that would seem the "correct" prefix.

smetj commented 8 years ago

yes, That sounds good