[jira] [Commented] (IMPALA-8845) Close ExecNode tree prior to calling FlushFinal in FragmentInstanceState

["Tim Armstrong (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IMPALA-8845?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16905370#comment-16905370 ] 

Tim Armstrong commented on IMPALA-8845:
---------------------------------------

I remember I had some concerns about doing an early close on a subtree that I documented here: IMPALA-3990. The concern was that the datastream sender might not tear itself down cleanly. I'm not sure if the code has changed since then.

The reason why we didn't see this problem was that we would only do an early close on a subtree either for peculiar queries or when the limit was hit at the coordinator, and the coordinator actually issues Cancel() RPCs to all the fragments.

My concern here would be that we might mess things up by relying on the Close() propagating down the tree instead of the coordinator sending out Cancel() RPCs.

> Close ExecNode tree prior to calling FlushFinal in FragmentInstanceState
> ------------------------------------------------------------------------
>
>                 Key: IMPALA-8845
>                 URL: https://issues.apache.org/jira/browse/IMPALA-8845
>             Project: IMPALA
>          Issue Type: Sub-task
>          Components: Backend
>            Reporter: Sahil Takiar
>            Assignee: Sahil Takiar
>            Priority: Major
>
> While testing IMPALA-8818, I found that IMPALA-8780 does not always cause all non-coordinator fragments to shutdown. In certain setups, TopN queries ({{select * from [table] order by [col] limit [limit]}}) where all results are successfully spooled, still keep non-coordinator fragments alive.
> The issue is that sometimes the {{DATASTREAM SINK}} for the TopN <-- Scan Node fragment ends up blocking waiting for a response to a {{TransmitData()}} RPC. This prevents the fragment from shutting down.
> I haven't traced the issue exactly, but what I *think* is happening is that the {{MERGING-EXCHANGE}} operator in the coordinator fragment hits {{eos}} whenever it has received enough rows to reach the limit defined in the query, which could occur before the {{DATASTREAM SINK}} sends all the rows from the TopN / Scan Node fragment.
> So the TopN / Scan Node fragments end up hanging until they are explicitly closed.
> The fix is to close the {{ExecNode}} tree in {{FragmentInstanceState}} as eagerly as possible. Moving the close call to before the call to {{DataSink::FlushFinal}} fixes the issue. It has the added benefit that it shuts down and releases all {{ExecNode}} resources as soon as it can. When result spooling is enabled, this is particularly important because {{FlushFinal}} might block until the consumer reads all rows.



--
This message was sent by Atlassian JIRA
(v7.6.14#76016)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscribe@impala.apache.org
For additional commands, e-mail: issues-all-help@impala.apache.org