Details
Description
This will help simplify the design of a lot of our internal code. The root tablet is really a special table, and should be treated specially.
Issue Links
- incorporates
-
ACCUMULO-335
Can't batchscan over the !METADATA table
-
- Resolved
-
- is duplicated by
-
ACCUMULO-1827
Broken Hadoop1 compatibility
-
- Resolved
-
- is related to
-
ACCUMULO-1998
Encrypted WALogs seem to be excessively buffering
-
- Resolved
-
-
ACCUMULO-2012
Cloning the metadata table fails with faulty error
-
- Resolved
-
- relates to
-
ACCUMULO-1143
unload of METADATA default_tablet with merging minor compactions can lockup
-
- Resolved
-
-
ACCUMULO-1651
GC removed WAL that master wasn't done with
-
- Resolved
-
- supercedes
-
ACCUMULO-1308
MakeDeleteEntries.call should only add entries if there are no !METADATA table references
-
- Resolved
-
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
Commit 1496171 from Christopher Tubbs
[ https://svn.apache.org/r1496171 ]
ACCUMULO-1481 Make root tablet its own table; further polishing and testing will be needed, but this changeset passes existing unit and integration build tests and most functional tests
Commit 1499510 from Christopher Tubbs
[ https://svn.apache.org/r1499510 ]
ACCUMULO-1481 : Add tests for splitting/merging root table; refactor to consolidate metadata constants and structures in an organized way; begin consolidating metadata ops into a servicer interface to abstract the code that actually does the servicing of metadata for all tables; remove some special cases for checking for metadata tables/root tablet that no longer apply; make mock work with root table; fix SimpleGarbageCollector to confirm candidates for deletion by checking the root table as well as the regular metadata table
Wait, what? Are we splitting and merging the root table now?
Are we splitting and merging the root table now?
No. You will get an error if you try to split the root table. You will get only a warning if you try to merge it (because merge on a single tablet is a NOP and there's no reason for this to fail).
The commit comment refers to the fact that I added a test to confirm this behavior.
Commit 4313860c41520c92cf2bc9709ac5fea91f94ecd4 in branch refs/heads/master from Eric Newton
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=4313860 ]
ACCUMULO-1481 remove warning from the shell when flushing !METADATA, since it no longer applies
Commit 6337c285fba031c0c14ab1a22e81e5c9b3c4c84c in branch refs/heads/master from Eric Newton
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=6337c28 ]
ACCUMULO-1481 create the root table entry in zookeeper, but blocked by ACCUMULO-1814
Commit e04fd1902b23b8bca8e1c30795f45d9c596f8623 in branch refs/heads/master from Eric Newton
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=e04fd19 ]
ACCUMULO-1481 creates root table and copies files to the directory on master start-up
Commit 7200e43a610694d2a9652d3d603bafc86e4395da in branch refs/heads/master from [~keith_turner]
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=7200e43 ]
ACCUMULO-1481 ACCUMULO-1823 fixed bug w/ major compaction
Commit c2dee4be5253b8c2f73a00068ab218b6ffc09bc8 in branch refs/heads/master from Eric Newton
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=c2dee4b ]
ACCUMULO-1481 broke binary compatibility with hadoop-1.2.0
Commit 1490795 from Christopher Tubbs
[ https://svn.apache.org/r1490795 ]
ACCUMULO-1481Use METADATA_TABLE_NAME constant wherever it wasn't already being used