Database Reference
In-Depth Information
During execution, the
hbck
tool scans the
.META.
table to gather all the information
it holds. It also scans the HDFS root directory that HBase is conigured to use. It then
compares the collected details to report on inconsistencies and integrity issues. This
tool checks the inconsistency at both region and table levels. The following are the
different types of checks:
•
Consistency check
: This check applies to a region. The tool checks whether
the region is listed in
.META.
and exists in HDFS. It also checks whether the
region should only be assigned to exactly one region server.
•
Integrity check
: This check is applicable to tables. As per this check any
rowkey should be assigned to a single region only.
This utility also provides the facility of ixing the inconsistencies. Some of the lags
used to ix the issues are as follows:
• Use
-fixAssignments
when:
° Regions are assigned to multiple RegionServers
° Regions are incorrectly assigned to a RegionServer but are being
served by some other RegionServer
° Regions exist in
.META.
but haven't been assigned to any
RegionServer
• Use
-fixHdfsOverlaps
when:
° Regions having overlapping key ranges
• Use
-fixHdfsOrphans
when:
° The
.regioninfo
file stored in HDFS that holds metadata for the
region is missing
HBase data import/export tools
HBase deals with lots of data spread over one or more regions and tables.
Sometimes, full or partial data needs to be either moved in or out from the cluster
either for data backup or data migration to another cluster. HBase comes with utility
tools for import and export data that use MapReduce jobs. These MapReduce jobs
can be used to write subsets, or an entire table, to lat iles in HDFS and also load
them back into the HBase cluster again.
