Recover Solr data from Fedora

[weiweishi]

We will need to figure out a way to reindex Solr from Fedora data in time of emergency recovery of data, if solr cloud redundancy and backups fail.

[mbarnett]

Notes for when it comes time to deal with this:

We can steal the basic sketch of how to do this from https://github.com/samvera/active_fedora/blob/7e9c365c00ced6ce4175096a3ff7b423cc72bf64/lib/active_fedora/indexing.rb#L95, although that code won't directly work for us because it doesn't account for the custom indexing and lambdas that come through from LockedLDPObject (it's not obvious to me that this works in raw Sufia either). Also, the "what Fedora objects get reindexed, what subobject graphs don't" code is rather convoluted: https://github.com/samvera/active_fedora/blob/e4685f953d3dcce8357d2e4e58df4092bbb52ea9/lib/active_fedora/indexing/descendant_fetcher.rb

Basic sketch of approach is:

1: Walk Fedora tree, looking for anything with a hasModel 2: Pull it via ActiveFedora::Base.find(id from uri of object from tree) (may need typecasting to concrete class?) (doesn't seem to, now that I've tried it) 3: Instantiate a LockedLDPObject and connect it's @ldp_object instance to the thing we pulled in 2. 4: Either unlock and save or directly write the solr data as in the first example above.

[pbinkley]

Any hope for doing all this from the triplestore and leave Fedora in peace? (that would assume that whatever broke Solr didn't also break the triplestore, of course).

[mbarnett]

I don't think so, at least not as long as we're dependent on ActiveFedora. It would be easier to decouple this stuff post-Fedora

[mbarnett]

Merged as of #809