Size of the index issues/issue can be quite huge when storing dozens of millions issues. Reducing its size allows to speed-up indexing time, especially the first time on fresh index, and to decrease IO consumption.
That is achieved by:
- dropping the unused fields attributes, gap and updatedAt
- loading issues from database instead of index in the web service batch/issues. That allows to:
- drop the fields message and manual_severity
- stop storing _source documents in the index.
At the time of writing, with Elasticsearch 2.4, these changes decrease size of index by 55%. The web service api/issues/search is also improved by stopping loading the whole documents just to get the issue keys.