Creating a new portal instance in a clustered environment from a non master node create index with incorrect mappings

• Type: Bug
• Status: Closed
• Resolution: Fixed
• Affects Version/s: 7.0.X
• Fix Version/s: 7.0.0 DXP FP86, 7.0.X
• Component/s: Core Infrastructure, Search Infrastructure
• Labels:
  • 7.0.x-only
  • ee-ts
  • liferay-fixpack-de-86-7010
• Branch Version/s:
  7.0.x
• Backported to Branch:
  Committed
• Fix Priority: 3
• Git Pull Request: https://github.com/brianchandotcom/liferay-portal-ee/pull/26417

When you create a new virtual instance from a non master node of a cluster index mapping are created incorrectly.

The listener in charge of create the index only runs from master node.

Steps to reproduce:

Start a two node cluster of Liferay.
Node 1 (first to start will be the master node of the cluster)
Node 2
Connect to a remote Elasticsearch for having clustered indexes.
Create a new virtual instance in node 2 (slave node)

In ElasticSearch logs we will observe this trace:

[INFO ][o.e.c.m.MetaDataCreateIndexService] [VLKs2wN] [liferay-XXXX] creating index, cause [auto(bulk api)], templates [], shards [5]/[1], mappings []

This is due to ElasticSearch is auto creating index instead of waiting for liferay CreateIndexRequest.

If we take a look at liferay-XXXXX index mappings with the next call:

_http://elasticsearch-ip:9200/liferay-*/_mapping

Mapping for lastName

Expected result:

"lastName":

{"type":"text","store":true}

,"lastName_sortable":

{"type":"keyword","store":true}

Current Result:

lastName":{"type":"text","fields":{"keyword":

{"type":"keyword","ignore_above":256}

}},"lastName_sortable":{"type":"text","fields":{"keyword":

{"type":"keyword","ignore_above":256}

}},
 

In 7.1 and above this issue doesn't occurs due to changes made by https://issues.liferay.com/browse/LPS-84409