[jira] Reopened: (HBASE-803) Atomic increment operations

["ryan rawson (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HBASE-803?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

ryan rawson reopened HBASE-803:
-------------------------------


the server code crashes if you increment during a flush, with an exception "more than 1 value from memcache.".  The solution is to only use the first value if there are more than 1 returned - with the snapshot, it's possible to return 2 values during a flush.

> Atomic increment operations
> ---------------------------
>
>                 Key: HBASE-803
>                 URL: https://issues.apache.org/jira/browse/HBASE-803
>             Project: Hadoop HBase
>          Issue Type: New Feature
>          Components: client, ipc, regionserver
>    Affects Versions: 0.19.0
>            Reporter: Jonathan Gray
>            Assignee: ryan rawson
>            Priority: Minor
>             Fix For: 0.19.1, 0.20.0
>
>         Attachments: HBASE-803-crash-bug.patch, hbase-803-v1.patch, HBASE-803-v2.patch, HBASE-803-v3-0.19.patch, HBASE-803-v3-0.20.patch, HBASE-803-v3.patch, HBASE-803.patch
>
>
> Discussion was already started in HBASE-798
> First version will contain to additional HTable methods:
> incrementColumnValue(table,row,family:column) -> Inserts a new version of the specified column (which must already exist) with an incremented cell value.
> incrementFamilyColumn(table,row,family: ) -> Inserts a new column under the specified family.  This assumes the family always contains only one column.  That column is read, incremented, deleted, and a new column is inserted with the incremented column name.  The existing value is re-inserted unchanged to the new column.
> I'm going to be making even more specialized versions of these for internal use (taking apart byte[]'s, doing different bit/byte operations, and assorted munging...).  So if anyone has any requests for other specialized versions, or more generic versions, please post them here and I'd be more than happy to take a stab at them.
> Next week I will likely have a go at HBASE-493

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