SOLR-13884: detect multiple replicas on single node

This commit is contained in:
yonik 2019-11-04 11:24:36 -05:00
parent 73c535261c
commit db65c82c39
1 changed files with 130 additions and 30 deletions

View File

@ -16,48 +16,31 @@
*/
package org.apache.solr.cloud.api.collections;
import java.io.IOException;
import java.lang.invoke.MethodHandles;
import java.nio.file.Path;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicReference;
import org.apache.lucene.util.LuceneTestCase.Nightly;
import org.apache.solr.SolrTestCaseJ4;
import org.apache.solr.client.solrj.SolrClient;
import org.apache.solr.client.solrj.SolrQuery;
import org.apache.solr.client.solrj.SolrRequest;
import org.apache.solr.client.solrj.embedded.JettySolrRunner;
import org.apache.solr.client.solrj.impl.CloudSolrClient;
import org.apache.solr.client.solrj.request.CollectionAdminRequest;
import org.apache.solr.client.solrj.request.UpdateRequest;
import org.apache.solr.client.solrj.response.CollectionAdminResponse;
import org.apache.solr.client.solrj.response.UpdateResponse;
import org.apache.solr.cloud.CloudTestUtils;
import org.apache.solr.cloud.MiniSolrCloudCluster;
import org.apache.solr.cloud.SolrCloudTestCase;
import org.apache.solr.common.SolrDocument;
import org.apache.solr.common.SolrDocumentList;
import org.apache.solr.common.cloud.ClusterState;
import org.apache.solr.common.cloud.DocCollection;
import org.apache.solr.common.cloud.Replica;
import org.apache.solr.common.util.IOUtils;
import org.apache.solr.common.util.TimeSource;
import org.apache.solr.util.TimeOut;
import org.apache.zookeeper.KeeperException;
import org.apache.solr.common.cloud.Slice;
import org.junit.After;
import org.junit.Before;
import org.junit.BeforeClass;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
public class ConcurrentCreateCollectionTest extends SolrCloudTestCase {
private static final Logger log = LoggerFactory.getLogger(MethodHandles.lookup().lookupClass());
@ -86,22 +69,94 @@ public class ConcurrentCreateCollectionTest extends SolrCloudTestCase {
}
private CollectionAdminRequest.Create createCollectionRequest(String cname, int numShards, int numReplicas) throws Exception {
CollectionAdminRequest.Create creq = CollectionAdminRequest
// nocommit .createCollection(cname, "conf", NODES - 1, NODES - 1)
.createCollection(cname, "conf", numShards, numReplicas)
.setMaxShardsPerNode(100);
creq.setWaitForFinalState(true);
creq.setAutoAddReplicas(true);
return creq;
}
public void testConcurrentCreatePlacement() throws Exception {
final int nThreads = 20;
final int nThreads = 2;
final int createsPerThread = 1;
final int repFactor = 1;
final boolean useClusterPolicy = true;
final boolean useCollectionPolicy = false;
final int nShards = 1;
final int repFactor = 2;
final boolean useClusterPolicy = false;
final boolean useCollectionPolicy = true;
final boolean startUnbalanced = true; // can help make a smaller test that can still reproduce an issue.
final int unbalancedSize = 1; // the number of replicas to create first
final boolean stopNode = false; // only applicable when startUnbalanced==true... stops a node during first collection creation, then restarts
final CloudSolrClient client = cluster.getSolrClient();
if (startUnbalanced) {
/*** This produces a failure (multiple replicas of single shard on same node) when run with NODES=4 and
final int nThreads = 2;
final int createsPerThread = 1;
final int nShards = 2;
final int repFactor = 2;
final boolean useClusterPolicy = false;
final boolean useCollectionPolicy = true;
final boolean startUnbalanced = true;
// NOTE: useClusterPolicy=true seems to fix it! So does putting both creates in a single thread!
// NOTE: even creating a single replica to start with causes failure later on.
Also reproduced with smaller cluster: NODES=2 and
final int nThreads = 2;
final int createsPerThread = 1;
final int nShards = 1;
final int repFactor = 2;
final boolean useClusterPolicy = false;
final boolean useCollectionPolicy = true;
final boolean startUnbalanced = true;
Also, with NODES=3:
final int nThreads = 2;
final int createsPerThread = 1;
final int nShards = 1;
final int repFactor = 2;
final boolean useClusterPolicy = false;
final boolean useCollectionPolicy = true;
final boolean startUnbalanced = false;
// Also succeeded in replicating a bug where all 5 replicas were on a single node: CORES=5, nThreads=5, repFactor=5,
// unbalancedSize = 16 (4 replicas on each of the up nodes), stopNode=true
***/
JettySolrRunner downJetty = cluster.getJettySolrRunners().get(0);
if (stopNode) {
cluster.stopJettySolrRunner(downJetty);
}
String cname = "STARTCOLLECTION";
CollectionAdminRequest.Create creq = CollectionAdminRequest
// nocommit .createCollection(cname, "conf", NODES - 1, NODES - 1)
.createCollection(cname, "conf", unbalancedSize, 1)
.setMaxShardsPerNode(100);
creq.setWaitForFinalState(true);
// creq.setAutoAddReplicas(true);
if (useCollectionPolicy) { creq.setPolicy("policy1"); }
creq.process(client);
if (stopNode) {
// this will start it with a new port.... does it matter?
cluster.startJettySolrRunner(downJetty);
}
}
if (useClusterPolicy) {
String setClusterPolicyCommand = "{" +
" 'set-cluster-policy': [" +
// " {'cores':'<100', 'node':'#ANY'}," +
" {'replica':'<2', 'shard': '#EACH', 'node': '#ANY'}," +
// " {'replica':'<2', 'node': '#ANY'}," +
" ]" +
"}";
@ -110,12 +165,36 @@ public class ConcurrentCreateCollectionTest extends SolrCloudTestCase {
}
if (useCollectionPolicy) {
// NOTE: the meer act of setting this named policy prevents LegacyAssignStrategy from being used, even if the policy is
// NOTE: the mere act of setting this named policy prevents LegacyAssignStrategy from being used, even if the policy is
// not used during collection creation.
String commands = "{set-policy :{policy1 : [{replica:'<2' , node:'#ANY'}]}}";
String commands = "{set-policy : {" +
" policy1 : [{replica:'<2' , node:'#ANY'}]" +
",policy2 : [{replica:'<2' , shard:'#EACH', node:'#ANY'}]" +
"}}";
client.request(CloudTestUtils.AutoScalingRequest.create(SolrRequest.METHOD.POST, commands));
/*** take defaults for cluster preferences
String cmd = "{" +
" 'set-cluster-preferences': [" +
// " {'cores':'<100', 'node':'#ANY'}," +
" {minimize:cores}" +
" ]" +
"}";
SolrRequest req = CloudTestUtils.AutoScalingRequest.create(SolrRequest.METHOD.POST, cmd);
client.request(req);
***/
}
/***
SolrRequest req = CloudTestUtils.AutoScalingRequest.create(SolrRequest.METHOD.GET, null);
SolrResponse response = req.process(client);
log.info("######### AUTOSCALE " + response);
***/
byte[] data = client.getZkStateReader().getZkClient().getData("/autoscaling.json", null, null, true);
log.info("AUTOSCALE DATA: " + new String(data, "UTF-8"));
final AtomicInteger collectionNum = new AtomicInteger();
Thread[] indexThreads = new Thread[nThreads];
@ -125,14 +204,17 @@ public class ConcurrentCreateCollectionTest extends SolrCloudTestCase {
try {
for (int j=0; j<createsPerThread; j++) {
int num = collectionNum.incrementAndGet();
// Thread.sleep(num*1000); // nocommit
String collectionName = "collection" + num;
CollectionAdminRequest.Create createReq = CollectionAdminRequest
.createCollection(collectionName, "conf", 1, repFactor)
.setMaxShardsPerNode(1);
.createCollection(collectionName, "conf", nShards, repFactor)
// .setMaxShardsPerNode(1) // should be default
;
createReq.setWaitForFinalState(false);
if (useCollectionPolicy) {
createReq.setPolicy("policy1");
}
createReq.setAutoAddReplicas(true);
createReq.process(client);
// cluster.waitForActiveCollection(collectionName, 1, repFactor);
@ -152,8 +234,9 @@ public class ConcurrentCreateCollectionTest extends SolrCloudTestCase {
thread.join();
}
int expectedTotalReplicas = nThreads * createsPerThread * repFactor;
int expectedTotalReplicas = unbalancedSize + nThreads * createsPerThread * nShards * repFactor;
int expectedPerNode = expectedTotalReplicas / NODES;
boolean expectBalanced = (expectedPerNode * NODES == expectedTotalReplicas);
Map<String,List<Replica>> replicaMap = new HashMap<>();
ClusterState cstate = client.getZkStateReader().getClusterState();
@ -173,11 +256,28 @@ public class ConcurrentCreateCollectionTest extends SolrCloudTestCase {
boolean failed = false;
for (List<Replica> replicas : replicaMap.values()) {
if (replicas.size() != expectedPerNode ) {
failed = true;
if (expectBalanced) {
failed = true;
}
log.error("UNBALANCED CLUSTER: expected replicas per node " + expectedPerNode + " but got " + replicas.size());
}
}
// check if there were multiple replicas of the same shard placed on the same node
for (DocCollection collection : cstate.getCollectionsMap().values()) {
for (Slice slice : collection.getSlices()) {
Map<String, Replica> nodeToReplica = new HashMap<>();
for (Replica replica : slice.getReplicas()) {
Replica prev = nodeToReplica.put(replica.getBaseUrl(), replica);
if (prev != null) {
failed = true;
// NOTE: with a replication factor > 2, this will print multiple times per bad slice.
log.error("MULTIPLE REPLICAS OF SINGLE SHARD ON SAME NODE: r1=" + prev + " r2=" + replica);
}
}
}
}
if (failed) {
log.error("Cluster state " + cstate.getCollectionsMap());
}