On the email@example.com mailing list, Yonik wrote:
"multiple values encountered for non multiValued field val_i: [10, 20]"
This should be very deterministic (i.e. it should always fail if it
were actually a non multiValued field).
The *_i fields are multivalued according to schema.xml, so this
exception should not happen (the version=1.0 in schema.xml means
multiValued=true by default).
Off of the top of my head, the only thing I can figure is that the
maven based tests are somehow getting the wrong schema sometimes.
Maybe if there's some different with how solr homes are set between
ant and maven?
Currently, under the Maven build, solr-core and solrj tests are run together under the solr-core module, because Maven can't handle the dependencies among solr-core, solr test-framework, and solrj. As a result, both solr-core and solrj resources are co-mingled. I think this is what's happening here: due to the apparently non-deterministic solr-home detection logic, in some environments, the "wrong" schema.xml file is used with some tests.