[
https://issues.apache.org/jira/browse/KUDU-3241?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17646294#comment-17646294
]
Alexey Serbin commented on KUDU-3241:
-------------------------------------
[~qijun],
It seems that might be
[KUDU-3406|https://issues.apache.org/jira/browse/KUDU-3406].
A manifestation of KUDU-3406 is that the OOM happens when CompactRowSetsOp is
run. Before the issue is fixed, as a workaround you could decrease
{{\-\-tablet_history_max_age_sec}} from its default setting of 604800 (became
that high staring Kudu 1.10.0) to some lower value, e.g. 900 (the default value
prior to 1.10.0 release).
> System under memory pressure (414.25% of limit used). However, there are no
> ops currently runnable which would free memory
> ---------------------------------------------------------------------------------------------------------------------------
>
> Key: KUDU-3241
> URL: https://issues.apache.org/jira/browse/KUDU-3241
> Project: Kudu
> Issue Type: Task
> Components: tserver
> Affects Versions: 1.10.0
> Environment: centos 7.7
> Reporter: Qijun Xie
> Priority: Major
> Fix For: n/a
>
>
> hi everybody，
> When I use kudu to store data in the production environment, I find that the
> memory of kudu tserver is not limited, or is constantly consuming the memory
> of the operating system, and I have not found an effective patch to fix this
> problem。
> One of my tserver, totally 256G memory, gflags:
> -memory_limit_hard_bytes= 26843545600 (25G) -memory_limit_soft_percentage=85
> -memory_pressure_percentage=80
> but in "top" and webui of tserver ,i find memory is used over 25G ,and then
> kill tserver because of OOM.
> could anybody give me some tips? thanks in advance.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
