[
https://issues.apache.org/jira/browse/AMBARI-14524?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dmytro Sen updated AMBARI-14524:
--------------------------------
Description:
PROBLEM: Ambari should give recommendations on this setting:
dfs.datanode.du.reserved
The default is 1GB and that is too low for a lot of our production customers.
If this is too low and the space on the datanode fills up, the nodemanagers
will start failing the health checks and we get into a bad state with not
enough nodemanagers being up. At this point the admin tries to run the balancer
and hope that it will balance very quickly.
There should be a recommendation on this value based on the disk space
available in the datanode volume. In addition, if the user sets it too low
there should be a warning saying this will cause problems if enough buffer is
not kept.
was:
d we get into a bad state with not enough nodemanagers being up. At this point
the admin tries to run the balancer and hope that it will balance very quickly.
There should be a recommendation on this value based on the disk space
available in the datanode volume. In addition, if the user sets it too low
there should be a warning saying this will cause problems if enough buffer is
not kept.
> HDFS Recommendation: dfs.datanode.du.reserved should be set to 10%-15% of
> volume size
> -------------------------------------------------------------------------------------
>
> Key: AMBARI-14524
> URL: https://issues.apache.org/jira/browse/AMBARI-14524
> Project: Ambari
> Issue Type: Bug
> Components: stacks
> Affects Versions: 2.1.0
> Reporter: Dmytro Sen
> Assignee: Dmytro Sen
> Priority: Critical
> Fix For: 2.4.0
>
>
> PROBLEM: Ambari should give recommendations on this setting:
> dfs.datanode.du.reserved
> The default is 1GB and that is too low for a lot of our production customers.
> If this is too low and the space on the datanode fills up, the nodemanagers
> will start failing the health checks and we get into a bad state with not
> enough nodemanagers being up. At this point the admin tries to run the
> balancer and hope that it will balance very quickly.
> There should be a recommendation on this value based on the disk space
> available in the datanode volume. In addition, if the user sets it too low
> there should be a warning saying this will cause problems if enough buffer is
> not kept.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
