Bug 1634036 - Per indice metrics missing in elasticsearch exporter
Summary: Per indice metrics missing in elasticsearch exporter
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: OpenShift Container Platform
Classification: Red Hat
Component: Logging
Version: 3.10.0
Hardware: x86_64
OS: Linux
high
medium
Target Milestone: ---
: 3.10.z
Assignee: Lukas Vlcek
QA Contact: Anping Li
URL:
Whiteboard:
Depends On:
Blocks: 1703118 1703136
TreeView+ depends on / blocked
 
Reported: 2018-09-28 14:01 UTC by Simon Reber
Modified: 2019-08-08 02:56 UTC (History)
8 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
: 1703118 (view as bug list)
Environment:
Last Closed: 2019-06-18 14:24:18 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Github vvanholl elasticsearch-prometheus-exporter pull 156 0 None None None 2019-03-06 09:05:23 UTC

Description Simon Reber 2018-09-28 14:01:41 UTC
Created attachment 1488081 [details]
Output from both command when reproducing the issue. Showing indices available in elasticsearch and metrics exported

Description of problem:

Running `atomic-openshift-3.10.14-1.git.0.ba8ae6d.el7.x86_64` with registry.access.redhat.com/openshift3/logging-elasticsearch:v3.10.14 is failing to export per indice metrics.

As per https://github.com/vvanholl/elasticsearch-prometheus-exporter the exporter should actually include per indice metrics, but as shown in the attached file no data related to the specific indices can be found

Version-Release number of selected component (if applicable):

 - atomic-openshift-3.10.14-1.git.0.ba8ae6d.el7.x86_64

How reproducible:

 - Always

Steps to Reproduce:
1. Install Prometheus according https://docs.openshift.com/container-platform/3.10/install_config/cluster_metrics.html#openshift-prometheus-deploy
2. Deploy EFK stack as per https://docs.openshift.com/container-platform/3.10/install_config/aggregate_logging.html#deploying-the-efk-stack
3. Check indices available in `elasticsearch` with `oc exec $es_pod -- curl -s -k --cert /etc/elasticsearch/secret/admin-cert --key /etc/elasticsearch/secret/admin-key https://logging-es:9200/_cat/indices?v`
4. Check Metrics reported by `elasticsearch` exporter: `curl -k -H "Authorization: Bearer $(oc sa get-token prometheus -n openshift-metrics)" https://<es-pod-ip>:4443/_prometheus/metrics`

Actual results:

No indice related metrics available/reported

Expected results:

Per indice metrics reported in order to use them further for charging, planning and monitoring purpose

Additional info:

Comment 1 Frederic Branczyk 2018-09-28 14:06:41 UTC
Changing this to the logging team as Prometheus just collects whatever the exporter gives it, and this seems to have shown already that the exporter is indeed not exporting the metrics.

Comment 2 Lukas Vlcek 2019-01-02 09:44:47 UTC
Indices level metrics are back in Prometheus exporter plugin. See https://github.com/vvanholl/elasticsearch-prometheus-exporter/pull/131
Next step is to upgrade to relevant ES plugin to have it in OpenShift logging.

Comment 12 Jeff Cantrill 2019-06-18 14:24:18 UTC
Closing CURRENTRELEASE as fixed in 3.11 and later.  Prior versions are based on earlier versions of Elasticsearch which we are moving away from.


Note You need to log in before you can comment on or make changes to this bug.