Summary

When pages have a significant number of previous versions, indexing will index all of them and cause delays with overall indexing.

Steps to Reproduce

1. Create a page with significant number of previous versions (in this case 340k)

Expected Results

1. Indexing will not be affected by the number of previous versions of a page and indexing will proceed normally

Actual Results

Indexing continues for a long periods of time depending on the number of historical versions and delay the index from completing for the entire instance.

1. Logs and DEBUG logging do not reveal explicit errors with index
2. Thread dumps show that most of the indexing threads are working on getVersionHistorySummary:

"Caesium-1-1" daemon prio=5 tid=0x00000000000000fa nid=0 runnable 
   java.lang.Thread.State: RUNNABLE
	at java.lang.Object.<init>(Object.java:37)
[...]
com.atlassian.confluence.core.persistence.hibernate.ContentEntityObjectHibernateDao.getVersionHistorySummary(ContentEntityObjectHibernateDao.java:344)

Notes

• Interested to know why historical versions of a page are included in the indexing process

Workaround

1. Delete the page with the large page history
2. Or make a copy of the page with a large history
   1. Then delete the original that has the history
   2. Rename the copy to remove "Copy" from the title