DataSource Creation reports Error in Console

Grafana Elasticsearch
1

With Grafana v4.02 running on RHEL 6.6 I am unable to add a Datasource to a new Gralog ElasticSearch Cluster which is a cluster.

I have an existing datasource to our Legacy Graylog host (not clustered) which works, however this is not behind a LoadBalancer. I have enabled debug logging for Grafana and when reproducing the problem (simply entering the https://loadbalncedIP:9000 and attempting to save the configuration and wild card for the index) i get in the console “ERROR”. However this does not report any thing in the logs.

If I attempt the connection to a single node on port 9000 (which is available tested with telnet). I do get a Info level error only reporting a 502 in the message.

Any thoughts as to why i get ERROR only and no other debug level information.

2

open chrome dev tools, any errors there? Maybe a CORS issue? Are you using direct or proxy access? If Proxy access can grafana-server access the loadbalancer?

3

Thanks for the response.
No Error reported in the browser (chrome) using the dev tools. I selected proxy access using https://lbal_IP:9000.

I am able to telnet to the load balanced IP which is listening on 9000, sending traffic on 9000 to the pool members. The servers are in different network segments so I do not believe there would be an issue source translation.

I have attached the payload in the post request

image
image760×314 10.4 KB

4

Testing this again. I did not see this before, there were some errors that were not noticed before.
Failed to load resource: the server responded with a status of 502 (Bad Gateway)
boot.85c49108.js:38 Objectconfig: Objectdata: ""headers: function ©status: 502statusText: "Bad Gateway"proto: Object
(anonymous) @ boot.85c49108.js:38

this is the URL that is being used which returns the 502
https:///api/datasources/proxy/14/grpmy7/31/2017og_11f7/31/20171ctor/_stats

This looking like an an issue on the elasticsearch endpoint?

5

looks like the loadbalancer is returning 502 http error

6

Thats what I though too, unfortunately the stack trace on the endpoint IP shows normal TCP com. and nothing that would indicate a reset on the part the load balancer.
I was able to recreate the 502 error and that was a true reset from the Lbal. Correcting my URL i know longer get the 502 however I still see this message. I believe this is from Graylog (Using Graylog Elastic Search Cluster). The version of graylog is 2.3.2, we have existing elastic seach data source from a graylog server with 2.0.2.
Here is the response:

Graylog Web Interface 
<script src="/assets/polyfill.35e28e4da7743596569a.js"></script>

<script src="/assets/plugin/org.graylog.plugins.pipelineprocessor.ProcessorPlugin/plugin.org.graylog.plugins.pipelineprocessor.PipelineProcessorPlugin.0cbe11128085ed745b77.js"></script>

<script src="/assets/plugin/org.graylog.plugins.map.MapWidgetPlugin/plugin.org.graylog.plugins.map.MapWidgetPlugin.5011f2ebff35ef956f1b.js"></script>

<script src="/assets/plugin/org.graylog.plugins.enterprise_integration.EnterpriseIntegrationPlugin/plugin.org.graylog.plugins.enterprise_integration.EnterpriseIntegrationPlugin.3e5ff59c400f972516bd.js"></script>

<script src="/assets/plugin/org.graylog.plugins.collector.CollectorPlugin/plugin.org.graylog.plugins.collector.CollectorPlugin.13edb0af11e27a17b001.js"></script>

<script src="/assets/app.35e28e4da7743596569a.js"></script>
7

looks like the url your using is not for an elasticsearch server but for a graylog server

8

Yep that was the issue… confused the API url endpoints between elasticsearch and graylog.