We recently migrated from Nuxeo 2017 to Nuxeo 2023, and performed a full reindex of all the 18m documents. When the reindex was completed, we discovered that not all documents have been indexed and according to "/nuxeo/api/v1/management/elasticsearch/checkSearch", 50.000 documents are missing.
Performing reindex on a single document on about 30 MB (20 versions) fails through the API and the bulk never completed. It does return an warning:
2025-02-20T11:38:44,118 WARN [bulk/indexPool-01,in:32,inCheckpoint:32,out:127,lastRead:1740046375389,lastTimer:0,wm:228071310162264065,loop:1463846,checkpoint] [org.nuxeo.elasticsearch.bulk.IndexRequestComputation] Indexing request for doc: 4b3c05fe-defd-4a6a-8b9a-510db3d58ef5, is very large: 212800287
When doing the reindex through Admin UI (JSF) an actual error is returned:
Caused by: org.nuxeo.ecm.core.api.NuxeoException: OpenSearchStatusException[Unable to parse response body]; nested: ResponseException[method [PUT], host [http://IP:9200], URI [/nuxeo/_doc/4b3c05fe-defd-4a6a-8b9a-510db3d58ef5?version_type=external&version=1740127768000276&timeout=120s], status line [HTTP/1.1 413 Request Entity Too Large]
The warning and error seem to be related to Opensearch, however, reuploading the file works without any problems and the file will be visible afterwards in Nuxeo UI, compared to files not indexed yet. Right now the theory is that it fails because of the 20 versions.
