HDDS-15102. Avoid ArchiveOutputStream.createArchiveEntry due to libnss issue

[symious]

What changes were proposed in this pull request?

We encountered DN randomly crash in our cluster, should be related to sssd bugs. Some detailed error logs can be found below.

This ticket is a workaround for this issue, so that sssd won't be used to trigger the crash of Datanode.

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007f2bb5f5dd2b, pid=2252464, tid=2472131
#
# JRE version: OpenJDK Runtime Environment (21.0.9) (build 21.0.9-internal-adhoc.root.jdk)
# Java VM: OpenJDK 64-Bit Server VM (21.0.9-internal-adhoc.root.jdk, mixed mode, sharing, tiered, compressed class ptrs, z gc, linux-amd64)
# Problematic frame:
# C  [libnss_sss.so.2+0x6d2b]  _nss_sss_endnetent+0x4b 

Stack: [0x00007f26e0bf6000,0x00007f26e0cf7000],  sp=0x00007f26e0cf4f10,  free space=1019k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C  [libnss_sss.so.2+0x6d2b]  _nss_sss_endnetent+0x4b
Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
J 35103  sun.nio.fs.UnixNativeDispatcher.getpwuid(I)[B java.base@21.0.9-internal (0 bytes) @ 0x00007f2ba229ae19 [0x00007f2ba229adc0+0x0000000000000059]
J 47972 c2 org.apache.hadoop.ozone.container.keyvalue.TarContainerPacker.includeFile(Ljava/io/File;Ljava/lang/String;Lorg/apache/commons/compress/archivers/ArchiveOutputStream;)V (80 bytes) @ 0x00007f2ba1401160 [0x00007f2ba14004c0+0x0000000000000ca0]
J 48138 c1 org.apache.hadoop.ozone.container.keyvalue.TarContainerPacker.pack(Lorg/apache/hadoop/ozone/container/common/interfaces/Container;Ljava/io/OutputStream;)V (105 bytes) @ 0x00007f2b98c6a3ac [0x00007f2b98c69e80+0x000000000000052c]
J 48133 c1 org.apache.hadoop.ozone.container.keyvalue.KeyValueContainer.packContainerToDestination(Ljava/io/OutputStream;Lorg/apache/hadoop/ozone/container/common/interfaces/ContainerPacker;)V (63 bytes) @ 0x00007f2b998a1aa4 [0x00007f2b998a17c0+0x00000000000002e4]
J 48131 c1 org.apache.hadoop.ozone.container.keyvalue.KeyValueContainer.exportContainerData(Ljava/io/OutputStream;Lorg/apache/hadoop/ozone/container/common/interfaces/ContainerPacker;)V (166 bytes) @ 0x00007f2b989993ec [0x00007f2b98998b60+0x000000000000088c]
J 48128 c1 org.apache.hadoop.ozone.container.replication.OnDemandContainerReplicationSource.copyData(JLjava/io/OutputStream;Lorg/apache/hadoop/ozone/container/replication/CopyContainerCompression;)V (58 bytes) @ 0x00007f2b98e71d04 [0x00007f2b98e70ea0+0x0000000000000e64]
J 48118 c1 org.apache.hadoop.ozone.container.replication.GrpcReplicationService.download(Lorg/apache/hadoop/hdds/protocol/datanode/proto/ContainerProtos$CopyContainerRequestProto;Lorg/apache/ratis/thirdparty/io/grpc/stub/StreamObserver;)V (296 bytes) @ 0x00007f2b999c4d14 [0x00007f2b999c4360+0x00000000000009b4]
J 48117 c1 org.apache.hadoop.hdds.protocol.datanode.proto.IntraDatanodeProtocolServiceGrpc$MethodHandlers.invoke(Ljava/lang/Object;Lorg/apache/ratis/thirdparty/io/grpc/stub/StreamObserver;)V (50 bytes) @ 0x00007f2b99314bcc [0x00007f2b993149e0+0x00000000000001ec]
J 35258 c2 org.apache.ratis.thirdparty.io.grpc.internal.ServerImpl$JumpToApplicationThreadServerStreamListener$1HalfClosed.runInContext()V (81 bytes) @ 0x00007f2ba0f58f18 [0x00007f2ba0f58d60+0x00000000000001b8]
J 16850 c2 org.apache.ratis.thirdparty.io.grpc.internal.ContextRunnable.run()V (35 bytes) @ 0x00007f2ba135e5c0 [0x00007f2ba135e220+0x00000000000003a0]
J 51312 c2 org.apache.ratis.thirdparty.io.grpc.internal.SerializingExecutor.run()V (114 bytes) @ 0x00007f2ba123dab4 [0x00007f2ba123d9a0+0x0000000000000114]
J 27132 c2 java.util.concurrent.ThreadPoolExecutor$Worker.run()V java.base@21.0.9-internal (9 bytes) @ 0x00007f2ba0fa2f24 [0x00007f2ba0fa2c80+0x00000000000002a4]
J 25894 c2 java.lang.Thread.run()V java.base@21.0.9-internal (23 bytes) @ 0x00007f2ba125b5cc [0x00007f2ba125b520+0x00000000000000ac]
v  ~StubRoutines::call_stub 0x00007f2b9fc52cbf 

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-15102

How was this patch tested?

This is a replacement of TarArchiveEntry initialization, existing unit test will be enough.

[adoroszlai]

Thanks @symious for the patch.

[adoroszlai]

Thanks @symious for updating the patch.

[adoroszlai]

I don't think setMode should be called. TarArchiveEntry(String) sets mode depending on whether name ends with /. It also sets linkFlag, which cannot be updated later. I think we should rely on this behavior by ensuring name ends with / for directories.

https://github.com/apache/commons-compress/blob/commons-compress-1.28.0/src/main/java/org/apache/commons/compress/archivers/tar/TarArchiveEntry.java#L616-L619

[symious]

Updated, PTAL.

[adoroszlai]

Thanks @symious for updating the patch. Looks like there are some related test failures:

https://github.com/apache/ozone/actions/runs/24875465109/job/72832265092?pr=10117#step:13:2514

https://github.com/apache/ozone/actions/runs/24875465109/job/72832265135?pr=10117#step:13:3765

https://github.com/apache/ozone/actions/runs/24875465109/job/72832265135?pr=10117#step:13:4039

[symious]

@adoroszlai Thanks for the reminder.

At first, it's only for includeFile, so the affected scope is smaller.

Let me try to fix these errors.

[symious]

@adoroszlai Updated, PTAL.

[adoroszlai]

Thanks @symious for the patch.

[symious]

@adoroszlai Thank you for the review and merge.