You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@beam.apache.org by "Beam JIRA Bot (Jira)" <ji...@apache.org> on 2021/03/25 17:20:00 UTC
[jira] [Commented] (BEAM-11626) Upgrade Guava to the latest version
except for Hadoop and Cassandra
[ https://issues.apache.org/jira/browse/BEAM-11626?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17308843#comment-17308843 ]
Beam JIRA Bot commented on BEAM-11626:
--------------------------------------
This issue is assigned but has not received an update in 30 days so it has been labeled "stale-assigned". If you are still working on the issue, please give an update and remove the label. If you are no longer working on the issue, please unassign so someone else may work on it. In 7 days the issue will be automatically unassigned.
> Upgrade Guava to the latest version except for Hadoop and Cassandra
> -------------------------------------------------------------------
>
> Key: BEAM-11626
> URL: https://issues.apache.org/jira/browse/BEAM-11626
> Project: Beam
> Issue Type: Improvement
> Components: build-system
> Reporter: Tomo Suzuki
> Assignee: Tomo Suzuki
> Priority: P2
> Labels: stale-assigned
> Attachments: compat_report.html
>
> Time Spent: 30h
> Remaining Estimate: 0h
>
> Related to https://issues.apache.org/jira/browse/BEAM-5559
> As per BEAM-8911, Cassandra-all requires Guava 25.1. But other parts of Beam dependencies (such as gcsio) may want to use newer version of Guava. Excerpt from Kiley's finding using Linkage Checker today:
> {noformat}
> (com.google.guava:guava:25.1-jre) com.google.common.util.concurrent.Uninterruptibles's method sleepUninterruptibly(java.time.Duration) is not found;
> referenced by 3 class files
> com.google.cloud.hadoop.gcsio.cooplock.CoopLockOperationDao (com.google.cloud.bigdataoss:gcsio:2.1.6)
> com.google.cloud.hadoop.gcsio.cooplock.CoopLockRecordsDao (com.google.cloud.bigdataoss:gcsio:2.1.6)
> com.google.cloud.hadoop.gcsio.testing.InMemoryObjectEntry (com.google.cloud.bigdataoss:gcsio:2.1.6)
> {noformat}
> Note that Guava's [Uninterruptibles.sleepUninterruptibly(Duration)|https://guava.dev/releases/30.1-jre/api/docs/com/google/common/util/concurrent/Uninterruptibles.html#sleepUninterruptibly(java.time.Duration)] has been introduced since 28.0.
> This ticket is attempts to declare the newer version of Guava while keeping the Guava version used by Beam's Cassandra tests.
> h1. Where is Guava dependency declared?
> {noformat}
> suztomo-macbookpro44% find . -name 'build.gradle' |xargs grep 'library.java.guava'
> ./sdks/java/core/build.gradle: shadowTest library.java.guava_testlib
> ./sdks/java/io/kinesis/build.gradle: compile library.java.guava
> ./sdks/java/io/kinesis/build.gradle: testCompile library.java.guava_testlib
> ./sdks/java/io/amazon-web-services2/build.gradle: testCompile library.java.guava_testlib
> ./sdks/java/io/google-cloud-platform/build.gradle: compile library.java.guava
> ./sdks/java/io/contextualtextio/build.gradle: testCompile library.java.guava_testlib
> ./sdks/java/extensions/sql/zetasql/build.gradle: compile library.java.guava
> ./sdks/java/maven-archetypes/examples/build.gradle: 'guava.version': dependencies.create(project.library.java.guava).getVersion(),
> ./runners/google-cloud-dataflow-java/build.gradle: testCompile library.java.guava_testlib
> {noformat}
> Other than tests, {{sdks/java/io/kinesis}}, {{sdks/java/io/google-cloud-platform}}, {{sdks/java/extensions/sql/zetasql}} will start declaring the new Guava dependency.
> The {{sdks/java/maven-archetypes/examples}} is tricky one. I want Hadoop/Cassandra users to use Guava 25.1 and others to use Guava 30.
> h1. Failures with Guava 30.1-jre
> Guava has removed methods and fields between 25.1-jre and 30.1-jre. The [API compliance checker|https://github.com/lvc/japi-compliance-checker#usage] report is attached: [^compat_report.html]
> With simply upgrading Guava version to 30.1-jre ([35f5b0|https://github.com/apache/beam/commit/35f5b0597e62563d843c9b3a7876212967b0a63c]), I get the following errors:
> https://gist.github.com/suztomo/f7bbf77b54265eb80fd3c2d230b678dc
> Failures in my local tests:
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemRegistrarTest and other tests in the package:
> {noformat}
> java.lang.NoSuchMethodError: com.google.common.util.concurrent.Futures.addCallback(Lcom/google/common/util/concurrent/ListenableFuture;Lcom/google/common/util/concurrent/FutureCallback;)V
> at org.apache.hadoop.hdfs.server.datanode.checker.ThrottledAsyncChecker.addResultCachingCallback(ThrottledAsyncChecker.java:167)
> at org.apache.hadoop.hdfs.server.datanode.checker.ThrottledAsyncChecker.schedule(ThrottledAsyncChecker.java:156)
> at org.apache.hadoop.hdfs.server.datanode.checker.StorageLocationChecker.check(StorageLocationChecker.java:165)
> at org.apache.hadoop.hdfs.server.datanode.DataNode.makeInstance(DataNode.java:2703)
> at org.apache.hadoop.hdfs.server.datanode.DataNode.instantiateDataNode(DataNode.java:2613)
> at org.apache.hadoop.hdfs.MiniDFSCluster.startDataNodes(MiniDFSCluster.java:1683)
> at org.apache.hadoop.hdfs.MiniDFSCluster.initMiniDFSCluster(MiniDFSCluster.java:905)
> at org.apache.hadoop.hdfs.MiniDFSCluster.<init>(MiniDFSCluster.java:517)
> at org.apache.hadoop.hdfs.MiniDFSCluster$Builder.build(MiniDFSCluster.java:476)
> at org.apache.beam.sdk.io.hdfs.HadoopFileSystemRegistrarTest.setUp(HadoopFileSystemRegistrarTest.java:58)
> at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> {noformat}
> This error appears in the following tests in sdks/java/io/hadoop-file-system module:
> {noformat}
> Test Result (22 failures / +22)
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemRegistrarTest.testServiceLoader
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testDeleteNonExisting
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testRenameExistingDestination
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testMatch
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testCopy
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testMatchForNonExistentFile
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testCreateAndReadFile
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testMatchDirectory
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testRenameRetryScenario
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testRenameMissingTargetDir
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testCreateAndReadFileWithShift
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testCreateAndReadFileWithShiftToEnd
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testCopySourceMissing
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testRenameMissingSource
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testMatchNewResource
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testMatchForRecursiveGlob
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testDelete
> org.apache.beam.sdk.io.hdfs.HadoopFileSystemTest.testRename
> org.apache.beam.sdk.io.hdfs.HadoopResourceIdTest.testGetFilename
> org.apache.beam.sdk.io.hdfs.HadoopResourceIdTest.testResourceIdTester
> {noformat}
> The deprecated [Futures.addCallback|https://guava.dev/releases/25.1-jre/api/docs/com/google/common/util/concurrent/Futures.html] has been removed since 26.0.
> and another error due to Guava version:
> {noformat}
> java.lang.RuntimeException: java.util.concurrent.ExecutionException: java.lang.NoSuchFieldError: DIGIT
> at org.apache.cassandra.utils.FBUtilities.waitOnFuture(FBUtilities.java:442)
> at org.apache.cassandra.db.SystemKeyspace.forceBlockingFlush(SystemKeyspace.java:829)
> at org.apache.cassandra.db.SystemKeyspace.removeTruncationRecord(SystemKeyspace.java:668)
> at org.apache.cassandra.db.ColumnFamilyStore.invalidate(ColumnFamilyStore.java:568)
> at org.apache.cassandra.db.ColumnFamilyStore.invalidate(ColumnFamilyStore.java:544)
> at org.apache.cassandra.schema.LegacySchemaMigrator.lambda$unloadLegacySchemaTables$1(LegacySchemaMigrator.java:137)
> at com.google.common.collect.ImmutableList.forEach(ImmutableList.java:406)
> at org.apache.cassandra.schema.LegacySchemaMigrator.unloadLegacySchemaTables(LegacySchemaMigrator.java:137)
> at org.apache.cassandra.schema.LegacySchemaMigrator.migrate(LegacySchemaMigrator.java:83)
> at org.apache.cassandra.service.CassandraDaemon.setup(CassandraDaemon.java:252)
> at org.apache.cassandra.service.CassandraDaemon.init(CassandraDaemon.java:513)
> at org.apache.cassandra.service.EmbeddedCassandraService.start(EmbeddedCassandraService.java:51)
> at org.apache.beam.sdk.io.hadoop.format.HadoopFormatIOCassandraTest.beforeClass(HadoopFormatIOCassandraTest.java:192)
> at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> {noformat}
> This error appear in this test in sdks/java/io/hadoop-format module:
> {noformat}
> org.apache.beam.sdk.io.hadoop.format.HadoopFormatIOCassandraTest.classMethod
> {noformat}
> This should be reproducible by {{./gradlew -p sdks/java/io/hadoop-format cleanTest test}}
> The {{sdks/java/io/cassandra}} module also had the problem:
> {noformat}
> ./gradlew sdks:java:io:cassandra:test --debug --stacktrace
> ...
> 2021-01-14T00:30:29.077-0500 [DEBUG] [TestEventLogger] Caused by: java.lang.NoSuchFieldError: DIGIT
> 2021-01-14T00:30:29.077-0500 [DEBUG] [TestEventLogger] at org.apache.cassandra.io.sstable.format.SSTableFormat$Type.<init>(SSTableFormat.java:63)
> 2021-01-14T00:30:29.077-0500 [DEBUG] [TestEventLogger] at org.apache.cassandra.io.sstable.format.SSTableFormat$Type.<clinit>(SSTableFormat.java:46)
> 2021-01-14T00:30:29.086-0500 [DEBUG] [TestEventLogger] at org.apache.cassandra.db.ColumnFamilyStore.getSSTablePath(ColumnFamilyStore.java:843)
> 2021-01-14T00:30:29.087-0500 [DEBUG] [TestEventLogger] at org.apache.cassandra.db.Memtable$FlushRunnable.<init>(Memtable.java:443)
> {noformat}
> Dependency graph of the hadoop-format module: https://gist.github.com/suztomo/8a444075d50e21cffab7bc9f087c82f3.
> h1. How about resolutionStrategy.force in the Hadoop modules?
> Kyle gave me advice about existing modules overriding version at module level, not BeamModulePlugin.groovy ([link|https://github.com/apache/beam/pull/13740#discussion_r562975904]).
>
> On ./sdks/java/io/hadoop-file-system module, this did not work:
> {code:groovy}
> configurations.testRuntime {
> resolutionStrategy {
> force 'com.google.guava:guava:25.1-jre'
> }
> }
> {code}
> This worked:
> {code:groovy}
> configurations.all {
> resolutionStrategy {
> force 'com.google.guava:guava:25.1-jre'
> }
> }
> {code}
> h1. Probably Irrelevant Cassandra Build Failure in my Linux host
> The build (master) fails in my Linux machine for Cassandra test:
> {noformat}
> suztomo@suztomo:~/beam$ ./gradlew :sdks:java:io:cassandra:test
> Configuration on demand is an incubating feature.
> > Task :sdks:java:io:cassandra:test
> #
> # A fatal error has been detected by the Java Runtime Environment:
> #
> # SIGSEGV (0xb) at pc=0x00007fbdbc875a3f, pid=3453267, tid=0x000000000034b177
> #
> # JRE version: OpenJDK Runtime Environment (8.0_181) (build 1.8.0_181-google-v7-341329124-341329124)
> # Java VM: OpenJDK 64-Bit Server VM (25.181-b01 mixed mode linux-amd64 compressed oops)
> # Problematic frame:
> # C [libc.so.6+0xbaa3f] __gettimeofday+0x7f
> #
> # Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
> ...
> {noformat}
> {{:sdks:java:io:cassandra:test}} works in my MacBook Pro. Other tests in {{:sdks:java:io}} passed in Linux.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)