This also brings up the question of what happens when the kernel version changes. Ideally, the correct version for the given time slice would be reported, even multiple ones if the kernel was updated.
Let me clarify first that we are not testing the system module in OpenSUSE. Said that, it's very unlikely that Metricbeat has a bug parsing kernel version that, by pure coincidence, produces a incorrect kernel version in the correct format.
As far as I know, kernel version is fetched and parsed on each "fetch" pass (ever X seconds) and it is not cached.
Are you sure you don't have a duplicated hostname, you are looking at the incorrect machine or anything like this? I'm open to think that we might have a bug, it's just that it seems unlikely that it's there.
@Mario_Castro Thanks for replying. I poked around in the configs, and I can't see what we'd have set up wrong that would have resulted in this discrepancy.
There are 4 hosts, each with their own kibana and ES, both clustered. The host name is clearly present in all queries and in metricbeat.
Btw, even if I select the last 1s of data in kibana's Metrics menu, and no data is showing because no such data has been collected, the kernel version is still shown there. So where does that data come from in that case?
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.
