Wildcard in index no longer works (used to) but specifying the index name works

[julsssark]

What happened:
Plug-in connected to a Graylog OpenSearch database and was working fine for months. The index field in the connection settings used a wildcard: opnsense_* . Now, when I click "Save & test" in the settings, an "Index not found: opnsense_*" error message is generated. If I change the index to "opnsense_8" (or any other valid index number) everything works fine.

What you expected to happen:
Using a wildcard for the index name does not generate an error on save and would allow for queries across all indexes that match the wildcard pattern.

How to reproduce it (as minimally and precisely as possible):

1. Navigate to Data Sources->Opensearch connection and open the settings
2. Confirm that index has a wildcard in it
3. Click "Save & test". Error should be displayed
4. Change wildcard to a number that has an existing index in the source
5. Click "Save & test". Successful message is displayed (Index OK. Time field name OK.)

Anything else we need to know?:
Running into the same error on another OpenSearch connection using a different index (also to Graylog)

Upgraded plug in to 2.21.3 and still see the same issue.

Environment:

• Grafana version: 11.3
• OpenSearch version: 2.17.1
• Plugin version: 2.21.3

[iwysiu]

Hi @julsssark ! Thanks for bug report, I've reproduced it and I'll move it to next for one of us to look at soon. We've recently refactored the health check, and we must have missed this case. The queries should still work if they used to, sorry for the inconvenience.

For whoever works on it: for the index opensearch_dashboards_* the old resource health check calls
<opensearchurl>/opensearch_dashboards_%2A/_mapping
and the new one calls <opensearchurl>/opensearch_dashboards_%2A/_mapping/field/<timefield>
I suspect that the field specific endpoint doesn't support the wildcard, which is why it used to work and doesn't now. We should make the healthcheck get all the fields like the old resource handler call and then parse them for the time field.

[julsssark]

@iwysiu Thanks for taking a look and confirming my bug report. I really like Grafana and I am happy to help out.

[yakir-shriker]

Same issue using Graylog with AWS Elasticsearch
Grafana version: 11.3
Elasticsearch version: 6.8
Plugin version: 2.21.3, tested also 2.21.1.

It was working before. We recently upgraded Grafana from 11 to 11.3, and I'm pretty sure that was the trigger for this issue.

[iwysiu]

Hi @julsssark and @yakir-shriker ! I just cut a release that reverts the health check code for now https://github.com/grafana/opensearch-datasource/releases/tag/v2.21.4

[julsssark]

Thank you @iwysiu! I can confirm that plugin 2.21.4 fixes the issue for me. I now can use the wildcard and "Save & Test" no longer generates an error.