[
https://issues.apache.org/jira/browse/HAWQ-883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15356563#comment-15356563
]
Ruilong Huo commented on HAWQ-883:
----------------------------------
[[email protected]], the GUC hawq_re_memory_overcommit_max is used to
define the maximum memory in MB that a hawq segment can overcommit during query
processing. It is 8192 MB by default.
gpcheck check hawq_re_memory_overcommit_max setting by comparing it with
physical memory on the node kmaster of your hawq cluster.
gpcheck complains that it has 3824 MB memory on kmaster, while
hawq_re_memory_overcommit_max is 8192 MB which is larger than physical memory.
As @lilima mentioned, either decrease hawq_re_memory_overcommit_max to the
value that is not larger than your physical memory, or increase your physical
memory on node would work.
> hawq check "hawq_re_memory_overcommit_max" error
> ------------------------------------------------
>
> Key: HAWQ-883
> URL: https://issues.apache.org/jira/browse/HAWQ-883
> Project: Apache HAWQ
> Issue Type: Bug
> Reporter: liuguo
> Assignee: Lei Chang
>
> [ERROR]:-host(kmaster): HAWQ master host memory size '3824' is less than the
> 'hawq_re_memory_overcommit_max' size '8192'
> When I set 'hawq_re_memory_overcommit_max=3000',then get an error:
> [ERROR]:-host(kmaster): HAWQ master's hawq_re_memory_overcommit_max GUC value
> is 3000, expected 8192
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
