Details
-
Bug
-
Status: Resolved
-
Minor
-
Resolution: Fixed
-
None
-
None
-
Reviewed
Description
Had a coworker try to use the RegionInfoMismatch tool I added in HBASE-26656. Curiously, the tool failed on the sanity check I added.
Aborting: sanity-check failed on updated RegionInfo. Expected encoded region name 736ee6186975de6967cd9e9e242423f0 but got 323748c77dde5b05982df0285b013232. Incorrectly created RegionInfo was: {ENCODED => 323748c77dde5b05982df0285b013232, NAME => 'test4,,1642405560420_0002.323748c77dde5b05982df0285b013232.', STARTKEY => '', ENDKEY => ''}
I couldn't understand why the tool wasn't working until I hooked up a debugger and realized that the problem wasn't in my code . The version of HBase on the system did not have the fix from HBASE-24500 included which meant that I was hitting the same "strange behavior", as Duo put it, in the RegionInfoBuilder "copy constructor".
While the versions of HBase which do not have this fix are EOL in terms of Apache releases, we can easily work around this in operator-tools (which may be used by any hbase 2.x release still in the wild).
Attachments
Issue Links
- relates to
-
HBASE-24500 The behavior of RegionInfoBuilder.newBuilder(RegionInfo) is strange
- Resolved
- links to