Elasticsearch purges may fail due to concurrent (or non-refreshed) writes

Description

Since the beginning of the week (approximately) we're having transient failures in IndexingActionInterceptorTest. The logs contain the following error:

See for instance http://ci.hibernate.org/job/hibernate-search-master/988/

I suspect it's a pre-existing bug, but we only see it now because of https://github.com/hibernate/hibernate-search/pull/1432, which had the effect of not executing subsequent works in a changeset when one work fails (while previously we considered each work independent from the others). (Note that this was still the right choice, since it aligned the behavior on embedded Lucene).

We should investigate why the delete by query fails in the first place.

Environment

None

Assignee

Yoann Rodière

Reporter

Yoann Rodière

Labels

None

Suitable for new contributors

None

Feedback Requested

None

Fix versions

Affects versions

Priority

Critical
Configure