[OAK-4697] Optimize read of old node state

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.5.9, 1.6.0
• Component/s: core, documentmk
• Labels: observation performance
• Epic Link: Improve DocumentMK performance
• Assignee: Marcel Reutegger
• Reporter: Marcel Reutegger
• Created: 24/Aug/16 11:43
• Updated: 15/Dec/16 15:21
• Resolved: 25/Aug/16 09:29

Description

Reading a node state with an old revision from a document can be expensive when many changes happened on a property in the meantime.

A typical stack trace looks like this:

	at org.apache.jackrabbit.oak.plugins.document.NodeDocument.getPreviousDocument(NodeDocument.java:1337)
	at org.apache.jackrabbit.oak.plugins.document.PropertyHistory$1.apply(PropertyHistory.java:70)
	at org.apache.jackrabbit.oak.plugins.document.PropertyHistory$1.apply(PropertyHistory.java:63)
	at com.google.common.collect.Iterators$8.transform(Iterators.java:794)
	at com.google.common.collect.TransformedIterator.next(TransformedIterator.java:48)
	at com.google.common.collect.Iterators$7.computeNext(Iterators.java:646)
	at com.google.common.collect.AbstractIterator.tryToComputeNext(AbstractIterator.java:143)
	at com.google.common.collect.AbstractIterator.hasNext(AbstractIterator.java:138)
	at com.google.common.collect.Iterators$PeekingImpl.hasNext(Iterators.java:1139)
	at org.apache.jackrabbit.oak.plugins.document.PropertyHistory$2.refillQueue(PropertyHistory.java:121)
	at org.apache.jackrabbit.oak.plugins.document.PropertyHistory$2.computeNext(PropertyHistory.java:96)
	at org.apache.jackrabbit.oak.plugins.document.PropertyHistory$2.computeNext(PropertyHistory.java:88)
	at com.google.common.collect.AbstractIterator.tryToComputeNext(AbstractIterator.java:143)
	at com.google.common.collect.AbstractIterator.hasNext(AbstractIterator.java:138)
	at org.apache.jackrabbit.oak.plugins.document.ValueMap$1$3.nextIterator(ValueMap.java:105)
	at org.apache.jackrabbit.oak.plugins.document.util.MergeSortedIterators.fetchNextIterator(MergeSortedIterators.java:98)
	at org.apache.jackrabbit.oak.plugins.document.util.MergeSortedIterators.next(MergeSortedIterators.java:85)
	at com.google.common.collect.Iterators$PeekingImpl.peek(Iterators.java:1162)
	at org.apache.jackrabbit.oak.plugins.document.util.MergeSortedIterators.adjustFirst(MergeSortedIterators.java:117)
	at org.apache.jackrabbit.oak.plugins.document.util.MergeSortedIterators.next(MergeSortedIterators.java:78)
	at org.apache.jackrabbit.oak.plugins.document.NodeDocument.getLatestValue(NodeDocument.java:1972)
	at org.apache.jackrabbit.oak.plugins.document.NodeDocument.getNodeAtRevision(NodeDocument.java:990)
	at org.apache.jackrabbit.oak.plugins.document.DocumentNodeStore.readNode(DocumentNodeStore.java:1079)

The read operation goes through the property history until it finds the most recent change. The old the read revision, the more changes are scanned.

Issue Links

• is related to: OAK-5207 Reduce reads in NodeDocument.getVisibleChanges()
• is related to: OAK-5206 Avoid merge sort of changes in intermediate split documents