Skip to content

actions: refactor delete indexes action to recheck rather than retry

Cwhite requested to merge T364190 into main

Why: The delete index action verifies deletion by rechecking the index api for presence of the deleted indexes. OpenSearch will respond 200 OK but acknowledged=false in the response body if the action is still being worked on after 30s. Curator has traditionally ignored this detail and would verify presence of the indexes regardless of what the server is doing and retries the delete upon finding the indexes aren't yet gone from the list indexes api. Retrying a delete while the delete is still completing yields 404 index_not_found_exception which bubbles the exception all the way up and ends the run.

What: Wait for indexes to disappear from the list index api rather than retry the delete request. We will leverage the existing timeout_override parameter to tell curator how long to wait.

Bug: T364190

Closes T364190

Merge request reports