I'm experiencing some strage behavior after upgrading to the super cool Logstash 5.2.0 + X-Pack. For the first 15 minutes after my cluster is brought up, I am able to get useful monitoring data about my Elasticsearch nodes. 15 minutes later, my nodes are all reporting "0" for most metrics, even though this seems purely cosmetic. The cluster is behaving correctly and the indexing rate has not dropped.
Here's a screenshot:
This only affect the Elasticsearch stats. I am able to get data about both Logstash and Kibana:
I have confirmed that this behavior only occurs with Logstash 5.2.0 by downgrading to 5.1.1 and monitoring the cluster for about an hour. Everything works as intended if I downgrade.
Hi @omarzing could you clarify the statement above? Did you install 5.2.0 Elasticsearch/Logstash/Kibana/X-Pack then saw that the Elasticsearch nodes were reporting 0 values oddly? Then you downgraded to 5.1.1 for all those things or just Logstash?
Did you install 5.2.0 Elasticsearch/Logstash/Kibana/X-Pack then saw that the Elasticsearch nodes were reporting 0 values oddly?
We are currently running a test cluster, so we had the luxury of reprovisioning from scratch. We initially reinstalled with the following:
Elasticsearch 5.2.0 (with X-Pack 5.2.0)
Kibana 5.2.0 (with X-Pack 5.2.)
Logstash 5.1.1
The cluster was running smoothly and monitoring data was unaffected with the above setup. We then took down our Logstash 5.1.1 nodes and reprovisioned Logstash 5.2.0 nodes with X-Pack 5.2.0. This is when we started seeing buggy stats.
Then you downgraded to 5.1.1 for all those things or just Logstash?
We destroyed our Logstash 5.2.0 boxes and then reprovisioned Logstash 5.1.1 boxes. Monitoring was back to normal.
We have since reinstalled Logstash 5.2.0 because the monitoring issue is not an absoulte show stopper for us.
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.