I have huge number of Issues in JIRA (6.X), is deindexing a good option

[Sridhar Nimmagadda]

I have a large number of Issues in JIRA close to a million !!

So instead of Archiving the issues, can i deindex old issues is this equivalent to archiving and also reducing load on the server where certain JQL's return a large number of issues (huge dataset)

Something similar to the below

IssueIndexManager indexManager = ComponentManager.getInstance().getIndexManager();
indexManager.deIndexIssueObjects(issues, true);

[Nic Brough -Adaptavist-]

Mmm. It'll shorten your searches, yes. But it might not be the ideal way to approach it.

An old site of mine did it differently - they changed the indexing *write* with their "archive rule". The reason they did it that way was that just de-indexing means the issues crawled back into the index if anyone stumbled onto an issue and updated it with anything. There's all sorts of paths to issues getting reindexed, and it does look very odd when you search for "all issues in project X", get 500 returned from the last 6 months and then 5 more scattered back over the predeeding 3 years. Or worse, you click "reindex"...

And, of course, de-indexing no matter what route you take, means you can't find old issues. If you can't find them, there's a strong argument to say you don't really need them and you should be simply archiving and deleting them (yes, I know, the argument against that is that Jira doesn't do archving. Yet. Have a look for "Blue Label" and Jira).

Edit - I forgot. That site ran into more problems later. Because they were missing index records for issues, they could not upgrade. One of the plugins they had needed to upgrade its data, but when it upgraded, it ran a search for the data belonging to it. No index record meant it couldn't find it, and hence didn't change it. So, when they decided to change the reindexing rule and let some data back in, the plugin choked on incompatible data. I suspect there are other gotchas to be had if you're dropping indices

[Sridhar Nimmagadda]

Yes i agree, de-index doesn't make sense for this situation.

Now does the JQL performance depend on the size of the Dataset it returns or the actual size of the indexes it has to search.

The typical suggestions for JQL's running slow has been about reducing the data size of (issues, projects, customfields....) the instance. Does a large instance resulting in large index size impact the search and ultimately in poor performance ?