Details
-
Bug Report
-
Resolution: Fixed
-
L3 - Default
-
3.8.0, 3.9.0-preview-1
-
None
-
Not defined
Description
Brief summary of the bug. What is it ? Where is it ?
During an investigation of an incident it was discovered that a third-party index shard was in a red state on Elasticsearch. This however caused the `readyz` endpoint of Optimize to return a 503 - service unavailable status code. The reason is that internally Optimize checks the overall clusterstate of elasticsearch and if it is `red` it reports itself as unavailable. This is a too broad scope of the check and needs adjustment to only check the state of the indices used by Optimize.
Steps to reproduce:
- Provoke a third-party index on elasticsearch to fail
- call the `/api/readyz` endpoint of Optimize
Actual result:
- observe a 503 status returned
Expected result:
- it should return 200, it may only return 503 if an actual optimize index is broken