8.4. HBase Known Issues

  • BUG-26592: Multiple tests are failing because of System time goes back frequently on SLES

    Problem: SUSE 11 SP3 clusters running on Amazon EC2 have had the system clock frequently going back (hundreds of ms), which causes multiple issues in HBase. The HBase data model assumes that timestamps are monotonically increasing, and failing that may result in updates being eclipsed by previous mutations.

    Workaround: Setup NTP and ensure that system clock time does not drift when running HBase. See https://hbase.apache.org/book.html#basic.prerequisites for more details.

  • BUG-23981 [Windows]: IOException: Invalid HFile block magic in Windows unit tests

    If you use HBase on a local filesystem (not HDFS) on Windows for single node non-distributed clusters, concurrent hfile readers might return data belonging to a different offset than what was requested, resulting in read failures.