[
https://issues.apache.org/jira/browse/HIVE-23236?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Marton Bod updated HIVE-23236:
------------------------------
Description:
Currently we have a global lock (NEXT_LOCK_ID) when running enqueueLock,
because the algorithm in checkLock requires the locks to have a well defined
order, and also requires that every lock component is already stored in the
RDBMS before checking the locks.
Proposed approach:
* Enqueue locks without a global S4U lock, using an auto-incremented value
instead to get the next lock ID (modify next_lock_id table)
* Before checking for lock conflicts, issue a S4U for all lock components we
are requesting (db/table/partition level) in order to prevent other clients for
doing interleaving conflict check (especially late-coming clients holding a
lower lockID)
* Conflict check algorithm to check not just for lower lockIDs, but also
higher lockIDs which are already in 'acquired' state
was:
Currently we have a global lock (NEXT_LOCK_ID) when running enqueueLock,
because the algorithm in checkLock requires the locks to have a well defined
order, and also requires that every lock component is already stored in the
RDBMS before checking the locks.
Proposed approach:
* Enqueue lock without a global lock, using a sequence for getting the next
lock ID
* Insert the locks as usual, but with an additional timestamp (enqueued_at)
* Check lock should check all enqueued locks for the given db/table/partition
but order should be determined based on the enqueued_at time instead of the
lockID
* If there is no blocking lock, update the state to acquired as usual
> Remove the global lock from acquireLock
> ---------------------------------------
>
> Key: HIVE-23236
> URL: https://issues.apache.org/jira/browse/HIVE-23236
> Project: Hive
> Issue Type: Improvement
> Reporter: Marton Bod
> Assignee: Marton Bod
> Priority: Major
>
> Currently we have a global lock (NEXT_LOCK_ID) when running enqueueLock,
> because the algorithm in checkLock requires the locks to have a well defined
> order, and also requires that every lock component is already stored in the
> RDBMS before checking the locks.
> Proposed approach:
> * Enqueue locks without a global S4U lock, using an auto-incremented value
> instead to get the next lock ID (modify next_lock_id table)
> * Before checking for lock conflicts, issue a S4U for all lock components we
> are requesting (db/table/partition level) in order to prevent other clients
> for doing interleaving conflict check (especially late-coming clients holding
> a lower lockID)
> * Conflict check algorithm to check not just for lower lockIDs, but also
> higher lockIDs which are already in 'acquired' state
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
