[
https://issues.apache.org/jira/browse/HIVE-5856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13903316#comment-13903316
]
Dave Winters commented on HIVE-5856:
------------------------------------
Last observed case was on HDP 1.3.2 in with Aster 5 client issuing large number
of calls from SQL-H software. Backend meta store is MySQL.
Seems to be mitigated by increasing the service stack and ram setting > 4Gb
Currently the status is being viewed as a possible memory leak issue that is
CND.
CND = can not duplicate
This situation pops up in very large tables with partitions (>200) and columns
(>300).
> Templeton service become nonresponsive
> --------------------------------------
>
> Key: HIVE-5856
> URL: https://issues.apache.org/jira/browse/HIVE-5856
> Project: Hive
> Issue Type: Bug
> Components: HCatalog
> Environment: HDP 1.3.x
> Reporter: Dave Winters
>
> The Templeton service stops responding to requests from clients.
> Service is running, metastore DB is running. Ports are in LISTEN state.
> All appears normal.
> Seems to occur at high request load rates.
> Current work around is to restart services.
> Still investigating and trying to capture one before reset by user/customer.
> Observed in SLES 11sp3 | CentOS and RHEL.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
