Solr init container does not update index configuration

[poikilotherm]

Since the merge of #175 for the salvation of #135, updating the index with the custom metadata blocks from Dataverse during startup does not work anymore.

[poikilotherm]

Reopening. This chicken and egg problem is not solvable as is now, as we need a reachable Dataverse for the index config and Dataverse needs a ready to serve Solr Index... (circular dependency)

[poikilotherm]

Idea: solve this via scripts and tools until the schema is managed by Dataverse...

1. The image will contain a standard schema as is now, but will contain include guards.
2. The image is extended by adding a startup script. If in an oppinionated (configurable) location a schema.xml is given and it is different to the one in place, replace it in our index and in the configset (so a new index will have the correct schema as of the first start)
3. Open question: how do we retrieve an updated schema when schemas are changed during runtime? sidecar? push/pull? restart pod? who and how do we trigger a reload?

• Make dvcli parse the TSV files and create the necessary <field>s and <copyField>s for us. (Needs more than just bash because of complex structure with multivalued attributes for compound fields)

[qqmyers]

Since Dataverse can start without solr's schema being updated (it only needs the right schema to index/reindex content), why do you consider this a circular dependency? Why won't the sequence of installing an updated Dataverse and then running an updating, using the Dataverse API to get the required fields to add to schema.xml, work?