8.3. Known Issues for HDP

  • BUG-15796: Sigsegv in mapred history server due to segfault in JniBasedUnixGroupsMapping.

    Problem: RHEL 6.2 and RHEL 6.3 contain known bugs in nslcd. On these platforms, we have observed instability and crashes in Hadoop daemons after an LDAP query issued by nslcd times out.

    Workaround: If your environment encounters this issue, then we recommend applying one of the following workarounds:

    • Upgrade to RHEL 6.4 or later.

    • Increase the configured nslcd search timeout in /etc/nslcd.conf.

    • Disable Hadoop native code integration for obtaining users' groups by setting hadoop.security.group.mappingtoorg.apache.hadoop.security.ShellBasedUnixGroupsMapping in core-site.xml.

  • BUG-825: EC2 m1.large cluster root partition is only 5GB and fills up quickly by HDP logs

    Problem: Directories and disks that you assign for logging in HDP do NOT have enough space to maintain logs during HDP operations.

    Workaround: Designate least 10 GB of free space on a disk that will be used by HDP logging.