@rmo-cdsp thanks for opening this issue! As we discussed at http://irclog.iq.harvard.edu/dataverse/2018-10-17 this information is not currently in http://guides.dataverse.org/en/4.9.4/admin/metadatacustomization.html

The file to edit is https://github.com/IQSS/dataverse/blob/develop/doc/sphinx-guides/source/admin/metadatacustomization.rst

For anyone picking this up, please see #5164 for examples of the struggles people have when trying to make even simple changes to the tsv files. We might want to include a troubleshooting section.

I mentioned at https://groups.google.com/d/msg/dataverse-community/9TNH41iGgG8/SZZaPbdvCwAJ that I was working on this issue and just created pull request #5251 and moved this to code review. Feedback is welcome!

@matthew-a-dunlap and I discussed how rather than documenting a fix we need for http://localhost:8080/api/admin/index/solr/schema to produce the right output ("_text_" instead of "text") that we should just fix it so I did. Everything else is docs and comments.

Looks good in general. Discussed some of the quirks with @pdurbin.

My to do list (will edit this comment for clarity in a bit):

• Clarify that TSVs are authortative and that the Google spreadsheet gets out of date. Done in d5fdd16.
• Add troubleshooting output for adding identifiers to controlled vocabulary values. Done in d9adf2b
• Additions from @qqmyers to the old doc: http://irclog.iq.harvard.edu/dataverse/2018-11-01#i_77561 (community tips section) . Done in 3bbde28.
• Document reloading. Done in 7765152.
• The order of fields by column (ie abovefold) doesn't matter. Done in d33b2aa.

Ok, as of d33b2aa I'm done hacking on this branch. I'm ready for more code review.

It would be easy to roll a fix for #3976 into the pull request if there's interest.

I addressed code review feedback in 2f5a1d6.

I'd like to explain the two code changes in the pull request.

First, in d9adf2b I documented an API for learning more about metadata fields which has existed since 4.0 (undocumented) and adjusted the API to provide information about controlled vocabulary values.

The second change is basically a bug fix. In the "Updating the Solr Schema" section of the main doc in the pull request I explain the curl command for getting Solr fields out of Dataverse to put into schema.xml. For Solr 4 the "catchall" field was text but for Solr 7 it's now _text_ so now the curl command output the correct field name.