[BUG] RFS Fails to read ES 5 created indices on ES 6 Snapshot #1247
Description
What is the bug?
A Elasticsearch 6 cluster may have indices originally created on ES 5 (e.g. via snapshot restore from an ES 5 snapshot). These indices will contain segments which were written on ES 5 (and thus Lucene 6) and may exist indefinitely on the cluster.
When RFS runs on the index, when it encounters segments written on ES 5, the following error occurs preventing the migration from proceeding
Caused by: java.lang.IllegalArgumentException: An SPI class of type org.apache.lucene.codecs.Codec with name 'Lucene62' does not exist. You need to add the corresponding JAR file supporting this SPI to your classpath. The current classpath supports the following names: [Lucene99, Lucene80, Lucene84, Lucene86, Lucene87, Lucene70, Lucene90, Lucene91, Lucene92, Lucene94, Lucene95]
at org.apache.lucene.util.NamedSPILoader.lookup(NamedSPILoader.java:113) ~[lucene-core-9.11.1.jar:9.11.1
What are your migration environments?
Source: ES 6.8, with index created on ES 5.6
Target: OS 2.17 (irrelevant)
How can one reproduce the bug?
Given a snapshot from an ES 5.x cluster, this can be restored onto an ES 6 cluster. Then a snapshot can be taken and reindex-from-snapshot can be initiated encountering the error.
An automated test reproducing the issue has been created here.
What is the expected behavior?
RFS reads the segments without issue.