[collectd] Problem with the way collect pushes JMX stats to Sensu

Joshua J. Kugler joshua at azariah.com
Sat Mar 19 20:33:32 CET 2016


[Sent this a couple days back. Never saw it show up on the mailing list or 
archives. Pardon if it's a dupe.]

Hello! I wrote a while back about trying to get the JMX plugin working, and 
the write_sensu plugin working.  The good news is, we have the JMX plugin 
working, and the write_sensu plugin is *kind of* working.  The issue we're 
having is how it reports the JMX data to sensu.

The full examples can be found here:
https://gist.github.com/jkugler/a0bb6e75cec9e946b2aa

However, the problem is this:

{
  "timestamp": "2016-03-18T00:07:50.872532+0000",
  "level": "info",
  "message": "publishing check result",
  "payload": {
    "client": "usw02-dev-dataservices-ts-cassandra-1",
    "check": {
      "name": "collectd", <--------

Every data point sent to Sensu has the same name, which means every data point 
from every bean shows up under "collectd" in the sensu dashbaord, so basically 
every bean is logged under the same "bag" of data, which means we can't 
extract any meaningful stats.

I see in the payload, there is this line:

      "collectd_plugin_instance": "thread_pools_misc_blocked",
also,in some cases:
      "collectd_plugin_type_instance": "livereadcount-real_leaders",


Is there a way to have it set those as the name? Or have Sensu key off that 
value for the categorization/separation of data??

Any insights or suggestions are appreciated!

j

-- 
Joshua J. Kugler - Fairbanks, Alaska
Azariah Enterprises - Programming and Website Design
joshua at azariah.com - Jabber: pedahzur at gmail.com
PGP Key: http://pgp.mit.edu/  ID 0x73B13B6A



More information about the collectd mailing list