On 10/29/2019 2:25 PM, Bridger Dyson-Smith wrote:
A very helpful person on IRC suggested that the only way to get rid of these might be a clean rebuild of the index, and that's not out of the question for us; I hoped to get a bit more information here.
I'm the one who you talked to on IRC.
Other than a clean reindexing of our data, is there anything we can do to clean these up? Thanks in advance for your help!
You should wait for confirmation, but I am not aware of any other way to fix this. The optimize operation (that I was hopeful would take care of it) is a purely Lucene operation that knows nothing at all about Solr. I learned that the optimize operation preserves all field metadata built into the index, even if the field was only referenced by deleted documents. Discussing the issue with other committers in our slack channel has revealed that it might be extremely difficult or impossible to improve the optimize operation so it purges unused metadata. I can ask on our dev list to see what I can learn.
I personally feel that Solr users should always be prepared to completely rebuild indexes from scratch. As painful as that prospect might be, it is the only solution to a number of problems, and is also frequently required by many configuration changes.
Thanks, Shawn
