[
https://issues.apache.org/jira/browse/HIVE-10632?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14630443#comment-14630443
]
Kristopher Kane commented on HIVE-10632:
----------------------------------------
In Hive .14 I am seeing this with the dropped table scenario. All existing
transactions are of state 'aborted' and
'select txn_id from TXNS where txn_id not in (select tc_txnid from
TXN_COMPONENTS) and txn_state = ‘a';' returns 0 rows since both tables contain
the same IDs.
Recreating the table had no effect on trigger a clean up process.
Is the only way to fix this to remove the entries from both transaction tables
in the RDBMS?
> Make sure TXN_COMPONENTS gets cleaned up if table is dropped before
> compaction.
> -------------------------------------------------------------------------------
>
> Key: HIVE-10632
> URL: https://issues.apache.org/jira/browse/HIVE-10632
> Project: Hive
> Issue Type: Bug
> Components: Metastore, Transactions
> Affects Versions: 1.0.0
> Reporter: Eugene Koifman
> Assignee: Alan Gates
>
> The compaction process will clean up entries in TXNS,
> COMPLETED_TXN_COMPONENTS, TXN_COMPONENTS. If the table/partition is dropped
> before compaction is complete there will be data left in these tables. Need
> to investigate if there are other situations where this may happen and
> address it.
> see HIVE-10595 for additional info
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)