honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Get max_seq_no after snapshot translog and Lucene (#45461) 

We should capture max_seq_no after snapshotting translog and Lucene;
otherwise, that max_seq_no can be smaller some operation in translog or
Lucene. With this change, we also hold the Engine#writeLock during this
check so that no indexing can happen.

Closes #45454
This commit is contained in:
Nhat Nguyen 2019-08-13 12:44:36 -04:00
parent adf8e20021
commit 24514275c7
1 changed files with 40 additions and 34 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

74
test/framework/src/main/java/org/elasticsearch/index/engine/EngineTestCase.java
View File

@ -68,6 +68,7 @@ import org.elasticsearch.common.settings.Settings;
import org.elasticsearch.common.unit.TimeValue; import org.elasticsearch.common.unit.TimeValue;
import org.elasticsearch.common.util.BigArrays; import org.elasticsearch.common.util.BigArrays;
import org.elasticsearch.common.util.set.Sets; import org.elasticsearch.common.util.set.Sets;
import org.elasticsearch.common.util.concurrent.ReleasableLock;
import org.elasticsearch.common.xcontent.NamedXContentRegistry; import org.elasticsearch.common.xcontent.NamedXContentRegistry;
import org.elasticsearch.common.xcontent.XContentBuilder; import org.elasticsearch.common.xcontent.XContentBuilder;
import org.elasticsearch.common.xcontent.XContentFactory; import org.elasticsearch.common.xcontent.XContentFactory;
@ -1075,42 +1076,47 @@ public abstract class EngineTestCase extends ESTestCase {
|| (engine instanceof InternalEngine) == false) { || (engine instanceof InternalEngine) == false) {
return; return;
} }
final long maxSeqNo = ((InternalEngine) engine).getLocalCheckpointTracker().getMaxSeqNo(); // hold writeLock during this check so indexing can't happen
final List<Translog.Operation> translogOps = new ArrayList<>(); try (ReleasableLock ignored = engine.writeLock.acquire()) {
try (Translog.Snapshot snapshot = EngineTestCase.getTranslog(engine).newSnapshot()) { final List<Translog.Operation> translogOps = new ArrayList<>();
Translog.Operation op; try (Translog.Snapshot snapshot = EngineTestCase.getTranslog(engine).newSnapshot()) {
while ((op = snapshot.next()) != null) { Translog.Operation op;
assertThat("translog operation [" + op + "] > max_seq_no[" + maxSeqNo + "]", op.seqNo(), lessThanOrEqualTo(maxSeqNo)); while ((op = snapshot.next()) != null) {
translogOps.add(op); translogOps.add(op);
}
}
final Map<Long, Translog.Operation> luceneOps = readAllOperationsInLucene(engine, mapper).stream()
.collect(Collectors.toMap(Translog.Operation::seqNo, Function.identity()));
for (Translog.Operation op : luceneOps.values()) {
assertThat("lucene operation [" + op + "] > max_seq_no[" + maxSeqNo + "]", op.seqNo(), lessThanOrEqualTo(maxSeqNo));
}
final long globalCheckpoint = EngineTestCase.getTranslog(engine).getLastSyncedGlobalCheckpoint();
final long retainedOps = engine.config().getIndexSettings().getSoftDeleteRetentionOperations();
final long seqNoForRecovery;
try (Engine.IndexCommitRef safeCommit = engine.acquireSafeIndexCommit()) {
seqNoForRecovery = Long.parseLong(safeCommit.getIndexCommit().getUserData().get(SequenceNumbers.LOCAL_CHECKPOINT_KEY)) + 1;
}
final long minSeqNoToRetain = Math.min(seqNoForRecovery, globalCheckpoint + 1 - retainedOps);
for (Translog.Operation translogOp : translogOps) {
final Translog.Operation luceneOp = luceneOps.get(translogOp.seqNo());
if (luceneOp == null) {
if (minSeqNoToRetain <= translogOp.seqNo()) {
fail("Operation not found seq# [" + translogOp.seqNo() + "], global checkpoint [" + globalCheckpoint + "], " +
"retention policy [" + retainedOps + "], maxSeqNo [" + maxSeqNo + "], translog op [" + translogOp + "]");
} else {
continue;
} }
} }
assertThat(luceneOp, notNullValue()); final Map<Long, Translog.Operation> luceneOps = readAllOperationsInLucene(engine, mapper).stream()
assertThat(luceneOp.toString(), luceneOp.primaryTerm(), equalTo(translogOp.primaryTerm())); .collect(Collectors.toMap(Translog.Operation::seqNo, Function.identity()));
assertThat(luceneOp.opType(), equalTo(translogOp.opType())); final long maxSeqNo = ((InternalEngine) engine).getLocalCheckpointTracker().getMaxSeqNo();
if (luceneOp.opType() == Translog.Operation.Type.INDEX) { for (Translog.Operation op : translogOps) {
assertThat(luceneOp.getSource().source, equalTo(translogOp.getSource().source)); assertThat("translog operation [" + op + "] > max_seq_no[" + maxSeqNo + "]", op.seqNo(), lessThanOrEqualTo(maxSeqNo));
}
for (Translog.Operation op : luceneOps.values()) {
assertThat("lucene operation [" + op + "] > max_seq_no[" + maxSeqNo + "]", op.seqNo(), lessThanOrEqualTo(maxSeqNo));
}
final long globalCheckpoint = EngineTestCase.getTranslog(engine).getLastSyncedGlobalCheckpoint();
final long retainedOps = engine.config().getIndexSettings().getSoftDeleteRetentionOperations();
final long seqNoForRecovery;
try (Engine.IndexCommitRef safeCommit = engine.acquireSafeIndexCommit()) {
seqNoForRecovery = Long.parseLong(safeCommit.getIndexCommit().getUserData().get(SequenceNumbers.LOCAL_CHECKPOINT_KEY)) + 1;
}
final long minSeqNoToRetain = Math.min(seqNoForRecovery, globalCheckpoint + 1 - retainedOps);
for (Translog.Operation translogOp : translogOps) {
final Translog.Operation luceneOp = luceneOps.get(translogOp.seqNo());
if (luceneOp == null) {
if (minSeqNoToRetain <= translogOp.seqNo()) {
fail("Operation not found seq# [" + translogOp.seqNo() + "], global checkpoint [" + globalCheckpoint + "], " +
"retention policy [" + retainedOps + "], maxSeqNo [" + maxSeqNo + "], translog op [" + translogOp + "]");
} else {
continue;
}
}
assertThat(luceneOp, notNullValue());
assertThat(luceneOp.toString(), luceneOp.primaryTerm(), equalTo(translogOp.primaryTerm()));
assertThat(luceneOp.opType(), equalTo(translogOp.opType()));
if (luceneOp.opType() == Translog.Operation.Type.INDEX) {
assertThat(luceneOp.getSource().source, equalTo(translogOp.getSource().source));
}
} }
} }
} }