Best techniques to clean up a big Jira instance

[adrien_rochereau]

Ok so over time we created lots of custom fields, some where deleted, some got a context added to make issues lighter,

however one thing I noticed, and that is well known, after adding a context to a field, the issue that should not have a value anymore are still present in the database.

in one case I have a project with 60k issues but 15 fields are only used on 30k issues,

how do I cleanup all the extra unusable/unwanted rows.

my main concern is that with the number of fields (around 900) the queries on the customfieldvalue table have become slower and during peak hours it is using a lot of our database processing.

I cannot find anything related to cleaning up tables or having jira do it either.

I assume a reindex is not gonna do anything as it is related to rebuilding the lucene index from the database.

Could I iterate with groovy on the issues and just do a issue save without modifying anything to force the database to clean?

[adrien_rochereau]

I am asking this one because customfieldvalue is the first table that is going to show in slow queries during crunch times, meaning if I can clean it, it will help the instance overall

[Matt Doar]

Generally its not the amount of data in a Jira database that causes performance problems. More important is the size of the Lucene indexes, particularly the main one (issues). A full reindex should reduce the size of your Lucene issues index if you have used custom field contexts to change the fields in each issue.

I don't recommend deleting rows from customfieldvalue without a full understanding of how other tables use it

[Matt Doar]

Though it is worth looking at the linked issue in https://community.atlassian.com/t5/Jira-questions/Jira-database-schema-change-for-customfieldvalue/qaq-p/743828#M241710