[infinispan-issues] [JBoss JIRA] (ISPN-11915) ServerRunMode.FORKED with maxIdle causes data loss

[Diego Lovison (Jira)]

     [ https://issues.redhat.com/browse/ISPN-11915?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Diego Lovison updated ISPN-11915:
---------------------------------
    Security Sensitive Issue: This issue is security relevant


> ServerRunMode.FORKED with maxIdle causes data loss
> --------------------------------------------------
>
>                 Key: ISPN-11915
>                 URL: https://issues.redhat.com/browse/ISPN-11915
>             Project: Infinispan
>          Issue Type: Bug
>    Affects Versions: 11.0.0.CR1, 10.1.8.Final
>            Reporter: Diego Lovison
>            Priority: Major
>         Attachments: test-log.txt
>
>
> With 3 servers in a cluster and maxIdle configured we can have data loss when the servers are in the same machine
> For master, just run the test
> {code:java}
> package org.infinispan.server.functional;
> import static org.junit.Assert.assertNotNull;
> import java.util.Arrays;
> import java.util.Collection;
> import java.util.UUID;
> import org.infinispan.client.hotrod.RemoteCache;
> import org.infinispan.configuration.cache.CacheMode;
> import org.infinispan.configuration.cache.ConfigurationBuilder;
> import org.infinispan.configuration.cache.StorageType;
> import org.infinispan.server.test.core.ServerRunMode;
> import org.infinispan.server.test.junit4.InfinispanServerRule;
> import org.infinispan.server.test.junit4.InfinispanServerRuleBuilder;
> import org.infinispan.server.test.junit4.InfinispanServerTestMethodRule;
> import org.infinispan.transaction.TransactionMode;
> import org.infinispan.util.concurrent.IsolationLevel;
> import org.junit.ClassRule;
> import org.junit.Rule;
> import org.junit.Test;
> import org.junit.runner.RunWith;
> import org.junit.runners.Parameterized;
> @RunWith(Parameterized.class)
> public class MaxIdleOperations {
>    @ClassRule
>    public static final InfinispanServerRule SERVERS =
>          InfinispanServerRuleBuilder.config("configuration/ClusteredServerTest.xml")
>                .runMode(ServerRunMode.FORKED)
>                .numServers(3)
>                .parallelStartup(false)
>                .build();
>    @Rule
>    public InfinispanServerTestMethodRule SERVER_TEST = new InfinispanServerTestMethodRule(SERVERS);
>    private final int maxIdle;
>    @Parameterized.Parameters(name = "{0}")
>    public static Collection<Object[]> data() {
>       // 70_000 is greater than runTestForMs
>       return Arrays.asList(new Object[][]{{100}, {1_000}, {10_000}, {70_000}});
>    }
>    public MaxIdleOperations(int maxIdle) {
>       this.maxIdle = maxIdle;
>    }
>    @Test
>    public void testMaxIdleOffHeapOperations() {
>       String cacheName = UUID.randomUUID().toString();
>       RemoteCache<String, String> remoteCache = SERVER_TEST.hotrod().create().getRemoteCacheManager()
>             .administration().getOrCreateCache(cacheName,
>             new ConfigurationBuilder()
>                   .clustering()
>                      .cacheMode(CacheMode.DIST_ASYNC).hash().numOwners(2).hash().numSegments(30)
>                   .locking()
>                      .isolationLevel(IsolationLevel.READ_COMMITTED).useLockStriping(false).lockAcquisitionTimeout(30000).concurrencyLevel(1000)
>                   .transaction()
>                      .transactionMode(TransactionMode.NON_TRANSACTIONAL)
>                   .memory()
>                      .storage(StorageType.OFF_HEAP).maxSize("400000")
>                   .expiration()
>                      .maxIdle(this.maxIdle).lifespan(300000)
>                   .persistence()
>                      .passivation(true).addSingleFileStore().maxEntries(1000000).shared(false).preload(false).fetchPersistentState(true).purgeOnStartup(true)
>             .build());
>       runTest(remoteCache);
>    }
>    @Test
>    public void testMaxIdleOperations() {
>       String cacheName = UUID.randomUUID().toString();
>       RemoteCache<String, String> remoteCache = SERVER_TEST.hotrod().create().getRemoteCacheManager()
>             .administration().getOrCreateCache(cacheName,
>             new ConfigurationBuilder()
>                   .clustering()
>                      .cacheMode(CacheMode.DIST_ASYNC).hash().numOwners(2).hash().numSegments(30)
>                   .locking()
>                      .isolationLevel(IsolationLevel.READ_COMMITTED).useLockStriping(false).lockAcquisitionTimeout(30000).concurrencyLevel(1000)
>                   .transaction()
>                      .transactionMode(TransactionMode.NON_TRANSACTIONAL)
>                   .expiration()
>                      .maxIdle(this.maxIdle).lifespan(300000)
>                   .persistence()
>                      .passivation(true).addSingleFileStore().maxEntries(1000000).shared(false).preload(false).fetchPersistentState(true).purgeOnStartup(true)
>             .build());
>       runTest(remoteCache);
>    }
>    private void runTest(RemoteCache<String, String> remoteCache) {
>       int runTestForMs = 30_000;
>       int maxKeys = 1000;
>       for (int i = 1; i <= maxKeys; i++) {
>          remoteCache.put(String.valueOf(i), "Test" + i);
>       }
>       long begin = System.currentTimeMillis();
>       long now = System.currentTimeMillis();
>       while (now - begin < runTestForMs) {
>          for (int i = 1; i <= maxKeys; i++) {
>             String key = String.valueOf(i);
>             String result = remoteCache.get(key);
>             String message = String.format("Null value for key: %s after %dms", key, now - begin);
>             assertNotNull(message, result);
>          }
>          now = System.currentTimeMillis();
>       }
>    }
> }
> {code}
> For 10.1.x, run the test with {{-Dorg.infinispan.test.server.dir=/path/to/infinispan-server-10.1.8.Final}}
> For embedded mode ( all servers are in the same JVM ) or when the server are running in a Container we don't have the issue.



--
This message was sent by Atlassian Jira
(v7.13.8#713008)