While this works pretty well, I discovered that some, not all, of my virtual machines change their recorded datacenter within the 10 seconds polling period.
and back with the next poll. I can assure you there is no vmotion going on in between. The VMs are not duplicate.
This applies to the virtualmachine metricset, host metricset is not affected, for the datastore metricset I cannot tell as we have the duplicate datastore names within the datacenters.
We are using vsphere Version 6.5.0 Build 4602587.
Metricbeat is running with version 6.0.1
There were recently quite some large changes to the vsphere module like https://github.com/elastic/beats/pull/4883 to fix some other issues. I wonder if this change potentially also fixed the problem above? This change did not make it into 6.1 but will be in 6.2 and is in master. Any chance you could try a snapshot build of metricbeat?
Well, that kind of solved the problem, as the field "vsphere.virtualmachine.datacenter.keyword" has vanished
I was interested in aggregated vsphere cluster statistics, the "datacenter" value was not what I actually needed, but I guess I can work with filters now
Thanks for the snapshot, this works for me now, at least I don't see inaccurate information currently.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.