Error after update from 1.7.5 to 2.3.2

I have a elasticsearch cluster running the 1.7.5 version. I've updated the cluster to Elasticsearch 2.3.2. The system works: Events are inserted, can query elastic search, but I see this error in the logs:

[2016-06-15 14:43:04,330][WARN ][index.warmer             ] [ip-10-0-0-222-avos] [logs-2016-06-15-23][0] failed to warm-up global ordinals for [message.has_alarm]
java.lang.ClassCastException: org.elasticsearch.index.fielddata.plain.PackedArrayIndexFieldData cannot be cast to org.elasticsearch.index.fielddata.IndexFieldData$Global
	at java.util.concurrent.ThreadPoolExecutor.runWorker(
	at java.util.concurrent.ThreadPoolExecutor$

This is a warning message that I can ignore or is better to investigate the issue. The problem is that this is an old index with eager_global_ordinals enabled in a boolean field. This can be done in the 1.7.5 but no in 2.3.2.

Ideally you should fix it by reindexing, but given it is a warning only you can probably live with it.