All Products
Search
Document Center

E-MapReduce:Known issues in JindoData 4.x

Last Updated:Jul 17, 2026

JindoData 4.x has known issues that may affect stability, performance, or data correctness. Review the issues for your version and apply the recommended fixes.

Version 4.6.x

Version 4.6.2

JindoSDK 4.6.0 and later versions support CRC64 checksum verification for write paths. By default, thefs.oss.checksum.crc64.enable parameter is enabled.

This configuration significantly reduces OSS-HDFS write performance. If performance is a priority, you can disable it. On the Configure tab of the Hadoop-Common service in the EMR Configure, add a configuration item in core-site.xml to set the fs.oss.checksum.crc64.enable parameter to false. For more information about how to add a configuration item, see Manage configuration items.

Version 4.6.1

  • In JindoSDK 4.6.1, when password-free access to OSS-HDFS is used in an EMR cluster, some jobs are interrupted while waiting for a token update.

    Use a fixed AccessKey or upgrade JindoSDK to version 4.6.2 or later. For upgrade instructions, see Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • In JindoSDK 4.6.1, using the JindoUtil tool with password-free access in an EMR cluster causes a permission error.

    Use a fixed AccessKey or upgrade JindoSDK to version 4.6.2 or later. For upgrade instructions, see Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • JindoSDK 4.6.0 and later versions support CRC64 checksum verification for write paths. By default, thefs.oss.checksum.crc64.enable parameter is enabled.

    This configuration significantly reduces OSS-HDFS write performance. If performance is a priority, you can disable it. On the Configure tab of the Hadoop-Common service in the EMR Configure, add a configuration item in core-site.xml to set the fs.oss.checksum.crc64.enable parameter to false. For more information about how to add a configuration item, see Manage configuration items.

Version 4.6.0

  • In JindoSDK 4.6.0, when password-free access to OSS-HDFS is used in an EMR cluster, some jobs hang while waiting for a token update.

    Use a fixed AccessKey or upgrade JindoSDK to version 4.6.2 or later. For upgrade instructions, see Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • In JindoSDK 4.6.0 and JindoFSx 4.6.0, setting fs.oss.credentials.provider=com.aliyun.jindodata.oss.auth.RangerCredentialsProvider in a Kerberos cluster causes a memory leak in the JindoFSx Namespace Service.

    Upgrade both JindoFSx and JindoSDK to version 4.6.2 or later. For more information, see Upgrade JindoData for an EMR cluster and Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • In JindoSDK 4.6.0, using the JindoUtil tool with password-free access in an EMR cluster causes a permission error.

    Use a fixed AccessKey or upgrade JindoSDK to version 4.6.2 or later. For upgrade instructions, see Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • JindoSDK 4.6.0 and later versions support CRC64 checksum verification for write paths. By default, thefs.oss.checksum.crc64.enable parameter is enabled.

    This configuration significantly reduces OSS-HDFS write performance. If performance is a priority, you can disable it. On the Configure tab of the Hadoop-Common service in the EMR Configure, add a configuration item in core-site.xml to set the fs.oss.checksum.crc64.enable parameter to false. For more information about how to add a configuration item, see Manage configuration items.

Version 4.5.x

Version 4.5.2

Version 4.5.1

Version 4.5.0

Version 4.4.x

Version 4.3.x

  • In JindoSDK 4.3.0 (on EMR-3.40.0 or EMR-5.6.0 clusters), displaying directory timestamps degrades ls command performance, so directory timestamps are not displayed.

    To display the timestamps, upgrade JindoSDK to version 4.3.1 or later. For upgrade instructions, see Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • In JindoSDK 4.3.0 (on EMR-3.40.0 or EMR-5.6.0 clusters), MagicCommitter can trigger frequent uploadPart calls, resulting in a "Part number must be an integer between 1 and 10000" error.

    To resolve this issue, upgrade JindoSDK to version 4.3.1 or later. For upgrade instructions, see Upgrade and roll back JindoSDK for an EMR cluster (X86).

  • JindoFSx 4.3.0 incorrectly handles errors when reading cached data from certain paths. The server does not return errors to the client, which then receives incorrect data.

    To resolve this issue, upgrade JindoFSx to version 4.3.1 or later. For more information, see Upgrade JindoData for an EMR cluster.

  • JindoFSx 4.3.0 incorrectly processes memory cache preload commands, which can corrupt cached data and cause subsequent reads to return incorrect results.

    To resolve this issue, upgrade JindoFSx to version 4.3.1 or later. For more information, see Upgrade JindoData for an EMR cluster.

  • JindoFSx 4.3.0 and 4.3.1 leak file handles. Over time, the process may reach the operating system's file handle limit, preventing the service from opening new handles and causing it to become unavailable.

    To resolve this issue, upgrade JindoFSx to version 4.3.2 or later. For more information, see Upgrade JindoData for an EMR cluster.

Version 4.2.x

In JindoSDK 4.2.0, the SEEK operation on large files can cause an integer overflow, which may cause SEEK-based jobs that read large files from OSS to fail.

Version 4.1.x

In JindoSDK 4.1.0, the SEEK operation on large files can cause an integer overflow, which may cause SEEK-based jobs that read large files from OSS to fail.

Version 4.0.x

In JindoSDK 4.0.0 (on EMR-3.39.0 or EMR-5.5.0 clusters), the SEEK operation on large files can cause an integer overflow, which may cause SEEK-based jobs that read large files from OSS to fail.

Limitations

  • JindoSDK does not support writing files that are larger than 80 GB to OSS.

  • JindoSDK does not support writing to OSS in append mode.

  • JindoSDK does not support client-side encryption for OSS.

  • JindoSDK does not support older versions of JindoFS in block mode and cache mode.

  • The Alibaba Cloud OSS-HDFS service (the JindoFS service) does not support upgrades from older versions of JindoFS running in block mode.

    You can use the JindoDistCp tool to migrate data from the old system to the new service.