[
https://issues.apache.org/jira/browse/HIVE-11388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14999937#comment-14999937
]
Eugene Koifman commented on HIVE-11388:
---------------------------------------
A simpler idea:
The Initiator places a row in COMPACTION_QUEUE to 'schedule' a compaction.
Before that it generates an ID from NEXT_COMPACTION_QUEUE_ID.
With HIVE-11948 we have support for SELECT FOR UPDATE on every DB except Derby,
though even w/o it we have Serializable isolation. Either way we should be
able to check that (db,table,partition,state=Initiated/Working) combination is
unique.
This allows Initiator to run on any/every metastore while ensuring there is no
double-compaction happening.
Need to come up with a solution for Cleaner. Could add another state, such as
'bc'= being cleaned, so that 1 cleaner marks a row as bc, and then others
ignore it. (This just needs to be cleaned up on failure)
This avoids the ZK.
> there should only be 1 Initiator for compactions per Hive installation
> ----------------------------------------------------------------------
>
> Key: HIVE-11388
> URL: https://issues.apache.org/jira/browse/HIVE-11388
> Project: Hive
> Issue Type: Bug
> Components: Transactions
> Affects Versions: 1.0.0
> Reporter: Eugene Koifman
> Assignee: Eugene Koifman
>
> org.apache.hadoop.hive.ql.txn.compactor.Initiator is a thread that runs
> inside the metastore service to manage compactions of ACID tables. There
> should be exactly 1 instance of this thread (even with multiple Thrift
> services).
> This is documented in
> https://cwiki.apache.org/confluence/display/Hive/Hive+Transactions#HiveTransactions-Configuration
> but not enforced.
> Should add enforcement, since more than 1 Initiator could cause concurrent
> attempts to compact the same table/partition - which will not work.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
