hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stack (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HBASE-623) migration script for hbase-82
Date Sun, 18 May 2008 22:01:55 GMT

     [ https://issues.apache.org/jira/browse/HBASE-623?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

stack updated HBASE-623:
------------------------

    Attachment: migrate.diff

Below is commit message.  Patch isn't ready yet... Verification scan needs to pause before
it runs for it to succeed.   Need to figure why.

{code}
HBASE-623 Migration for HBASE-82

M  src/test/org/apache/hadoop/hbase/TestSerialization.java
    (testHMsg): Renamed from testname.
M  src/test/org/apache/hadoop/hbase/TestClassMigration.java
    Test for self-migration of classes.  Has test for HStoreKey.
M  src/test/org/apache/hadoop/hbase/regionserver/TestHStoreFile.java
    readSplitInfo was moved into HStoreFile
M  src/test/org/apache/hadoop/hbase/util/TestMigrate.java
    (verify): Verify that we can actually read the migrated
    table and that the migrated table has expected content in it.
M  src/java/org/apache/hadoop/hbase/HColumnDescriptor.java
    Add check for colon on family names.  HCDs family names no 
    longer have the colon.  We need this check for case where
    we are reading in data written with a previous hbase version.
M  src/java/org/apache/hadoop/hbase/HConstants.java
    Added defines for hbase.client.retries.number
M  src/java/org/apache/hadoop/hbase/regionserver/HStoreFile.java
    Moved readSplitInfo from HStore to here.  Also changed the
    way we serialize back to writing the encoded region name
    as a String rather than as an int; the encoded name is written
    into the info file that accompanies HSFs.  Staying on String
    means we do not have to write a migration for old info files
    so new code can read them.
M src/java/org/apache/hadoop/hbase/regionserver/HStore.java
    Moved readSplitInfo into HStoreFile.
M src/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java
    Fixed a debug log.
M src/java/org/apache/hadoop/hbase/util/Bytes.java
    Use encoded lengths as Text does.  Makes it so we don't have to
    migrate old data; our byte buffer reads will just work.
M  src/java/org/apache/hadoop/hbase/client/HTable.java
    Added getScanner overrides.
M src/java/org/apache/hadoop/hbase/client/HConnectionManager.java
    Made logging a little more informative.  Removed sleeps even when
    no retries left. Made the master not running exception a little more 
    informative.
{code}

> migration script for hbase-82
> -----------------------------
>
>                 Key: HBASE-623
>                 URL: https://issues.apache.org/jira/browse/HBASE-623
>             Project: Hadoop HBase
>          Issue Type: Bug
>            Reporter: stack
>            Assignee: stack
>            Priority: Blocker
>             Fix For: 0.2.0
>
>         Attachments: migrate.diff
>
>
> When hbase-82 goes in, will need a migration script for pre-hbase-82 installs.  I don't
think its going to be as bad as originally thought -- i.e. rewriting all data -- but this
JIRA is about making sure old data comes across w/o issue.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message