org.elasticsearch.index.shard.ElasticsearchMergePolicy Maven / Gradle / Ivy
Show all versions of elasticsearch Show documentation
/*
* Licensed to Elasticsearch under one or more contributor
* license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright
* ownership. Elasticsearch licenses this file to you under
* the Apache License, Version 2.0 (the "License"); you may
* not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
package org.elasticsearch.index.shard;
import org.apache.logging.log4j.LogManager;
import org.apache.logging.log4j.Logger;
import org.apache.lucene.index.FilterMergePolicy;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.MergePolicy;
import org.apache.lucene.index.SegmentCommitInfo;
import org.apache.lucene.index.SegmentInfos;
import org.elasticsearch.Version;
import java.io.IOException;
import java.util.Collections;
import java.util.Map;
/**
* A {@link MergePolicy} that upgrades segments and can upgrade merges.
*
* It can be useful to use the background merging process to upgrade segments,
* for example when we perform internal changes that imply different index
* options or when a user modifies his mapping in non-breaking ways: we could
* imagine using this merge policy to be able to add doc values to fields after
* the fact or on the opposite to remove them.
*
* For now, this {@link MergePolicy} takes care of moving versions that used to
* be stored as payloads to numeric doc values.
*/
public final class ElasticsearchMergePolicy extends FilterMergePolicy {
private static final Logger logger = LogManager.getLogger(ElasticsearchMergePolicy.class);
// True if the next merge request should do segment upgrades:
private volatile boolean upgradeInProgress;
// True if the next merge request should only upgrade ancient (an older Lucene major version than current) segments;
private volatile boolean upgradeOnlyAncientSegments;
private static final int MAX_CONCURRENT_UPGRADE_MERGES = 5;
/** @param delegate the merge policy to wrap */
public ElasticsearchMergePolicy(MergePolicy delegate) {
super(delegate);
}
/** return the wrapped merge policy */
public MergePolicy getDelegate() {
return in;
}
private boolean shouldUpgrade(SegmentCommitInfo info) {
org.apache.lucene.util.Version old = info.info.getVersion();
org.apache.lucene.util.Version cur = Version.CURRENT.luceneVersion;
// Something seriously wrong if this trips:
assert old.major <= cur.major;
if (cur.major > old.major) {
// Always upgrade segment if Lucene's major version is too old
return true;
}
if (upgradeOnlyAncientSegments == false && cur.minor > old.minor) {
// If it's only a minor version difference, and we are not upgrading only ancient segments,
// also upgrade:
return true;
}
// Version matches, or segment is not ancient and we are only upgrading ancient segments:
return false;
}
@Override
public MergeSpecification findForcedMerges(SegmentInfos segmentInfos,
int maxSegmentCount, Map segmentsToMerge, MergeContext mergeContext)
throws IOException {
if (upgradeInProgress) {
MergeSpecification spec = new MergeSpecification();
for (SegmentCommitInfo info : segmentInfos) {
if (shouldUpgrade(info)) {
// TODO: Use IndexUpgradeMergePolicy instead. We should be comparing codecs,
// for now we just assume every minor upgrade has a new format.
logger.debug("Adding segment {} to be upgraded", info.info.name);
spec.add(new OneMerge(Collections.singletonList(info)));
}
// TODO: we could check IndexWriter.getMergingSegments and avoid adding merges that IW will just reject?
if (spec.merges.size() == MAX_CONCURRENT_UPGRADE_MERGES) {
// hit our max upgrades, so return the spec. we will get a cascaded call to continue.
logger.debug("Returning {} merges for upgrade", spec.merges.size());
return spec;
}
}
// We must have less than our max upgrade merges, so the next return will be our last in upgrading mode.
if (spec.merges.isEmpty() == false) {
logger.debug("Returning {} merges for end of upgrade", spec.merges.size());
return spec;
}
// Only set this once there are 0 segments needing upgrading, because when we return a
// spec, IndexWriter may (silently!) reject that merge if some of the segments we asked
// to be merged were already being (naturally) merged:
upgradeInProgress = false;
// fall through, so when we don't have any segments to upgrade, the delegate policy
// has a chance to decide what to do (e.g. collapse the segments to satisfy maxSegmentCount)
}
return super.findForcedMerges(segmentInfos, maxSegmentCount, segmentsToMerge, mergeContext);
}
/**
* When upgrade is true, running a force merge will upgrade any segments written
* with older versions. This will apply to the next call to
* {@link IndexWriter#forceMerge} that is handled by this {@link MergePolicy}, as well as
* cascading calls made by {@link IndexWriter}.
*/
public void setUpgradeInProgress(boolean upgrade, boolean onlyAncientSegments) {
this.upgradeInProgress = upgrade;
this.upgradeOnlyAncientSegments = onlyAncientSegments;
}
}