You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Sahil Takiar (JIRA)" <ji...@apache.org> on 2018/02/08 01:07:00 UTC
[jira] [Updated] (HIVE-17837) Explicitly check if the HoS Remote
Driver has been lost in the RemoteSparkJobMonitor
[ https://issues.apache.org/jira/browse/HIVE-17837?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sahil Takiar updated HIVE-17837:
--------------------------------
Attachment: HIVE-17837.2.patch
> Explicitly check if the HoS Remote Driver has been lost in the RemoteSparkJobMonitor
> -------------------------------------------------------------------------------------
>
> Key: HIVE-17837
> URL: https://issues.apache.org/jira/browse/HIVE-17837
> Project: Hive
> Issue Type: Sub-task
> Components: Hive
> Reporter: Sahil Takiar
> Assignee: Sahil Takiar
> Priority: Major
> Attachments: HIVE-17837.1.patch, HIVE-17837.2.patch
>
>
> Right now the {{RemoteSparkJobMonitor}} implicitly checks if the connection to the Spark remote driver is active. It does this everytime it triggers an invocation of the {{Rpc#call}} method (so any call to {{SparkClient#run}}).
> There are scenarios where we have seen the {{RemoteSparkJobMonitor}} when the connection to the driver dies, because the implicit call fails to be invoked (see HIVE-15860).
> It would be ideal if we made this call explicit, so we fail as soon as we know that the connection to the driver has died.
> The fix has the added benefit that it allows us to fail faster in the case where the {{RemoteSparkJobMonitor}} is in the QUEUED / SENT state. If its stuck in that state, it won't fail until it hits the monitor timeout (by default 1 minute), even though we already know the connection has died. The error message that is thrown is also a little imprecise, it says there could be queue contention, even though we know the real reason is that the connection was lost.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)