During an upgrade from Jira 3.13.4 to 4.2.2, I couldn't help but notice that issue re-indexing was carried out 4/5 times
The first re-index is carried out immediately after data import, then a number of upgrade tasks run, 3 or 4 of which carry out issue re-index again.
Our system has 200,000+ issues in it, and each index job takes approx 90-100 minutes.
Perhaps all of the re-index jobs are needed, but I can't think of a reason why any more than one re-index would be needed?
Can Atlassian insert some intelligence into the upgrade process to remove all unnecessary re-index steps?