HBCK: What It Checks, Common Commands, and Repair Strategies for HBase

HBCK is a diagnostic utility for HBase that primarily checks two aspects: (1) region consistency – ensuring every region is assigned, deployed on a single RegionServer, and that its state is consistent in memory, the hbase:meta table, and ZooKeeper; and (2) table integrity – guaranteeing each rowkey resides in only one region interval.

Common hbck commands include:

./bin/hbase hbck

./bin/hbase hbck --details

./bin/hbase hbck TableFoo TableBar

Low‑risk repair options:

-fixAssignments – fixes regions that are unassigned, incorrectly assigned, or assigned to multiple RegionServers.

-fixMeta – reconciles discrepancies between .regioninfo files on HDFS and entries in the hbase:meta table, using HDFS as the source of truth.

High‑risk repairs involve fixing overlapping region intervals, which may require direct HDFS file modifications or manual intervention; such operations should be preceded by hbck --details to understand the issue, and the -repair or -fix commands are strongly discouraged in production environments.

Typical RIT (Region‑in‑Transaction) handling patterns:

Pending open/close regions can often be repaired with hbck.

Failed open/close regions usually cannot be fixed by hbck and need log investigation.

For failed open/close regions, examine logs to determine the root cause.

If a region is in RIT but hbck reports normal, delete the corresponding node under /zookeeper/quorum/region-in-transaction and restart the master.

Log analysis is essential for pinpointing the exact cause of HBase issues, as monitoring only provides indirect clues. By examining logs, administrators can trace the precise sequence of events leading to a problem.

The article also includes several illustrative screenshots (omitted here) and references a “NetEase HBase Practice” meetup presentation for deeper insight into HBase source‑code debugging.