hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Reid Chan (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HDFS-11760) Make CHECK_CANCEL_INTERVAL configurable
Date Fri, 05 May 2017 06:06:04 GMT

     [ https://issues.apache.org/jira/browse/HDFS-11760?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Reid Chan updated HDFS-11760:
-----------------------------
    Description: 
There's a case in my company: when two namenodes tried to transit state, checkpoint took place
in standbynode and it didn't check cancellation until every after 4096 inodes saved, which
took too much time between checks in our case and caused state transition fail at the end.
After we changed it to 50, state transition succeed.

The *CHECK_CANCEL_INTERVAL* was introduced in HDFS-7097 and set to 4096, previously was 50.
There may be some reasons for the choice of 4096, but i'm not sure that, so I suggest to make
it configurable.

  was:
There's a case in my company: when two namenodes tried to transit state, checkpoint took place
in standbynode and it didn't check cancellation until every after 4096 inodes saved, which
took too much time between checks in our case and caused state transition fail at the end.
After we changed it to 50, state transition succeed.

The {quote}CHECK_CANCEL_INTERVAL{quote} was introduced in HDFS-7097 and set to 4096, previously
was 50. There may be some reasons for the choice of 4096, but i'm not sure that, so I suggest
to make it configurable.


> Make CHECK_CANCEL_INTERVAL configurable
> ---------------------------------------
>
>                 Key: HDFS-11760
>                 URL: https://issues.apache.org/jira/browse/HDFS-11760
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>    Affects Versions: 2.7.0, 2.7.1, 2.7.2, 2.7.3, 3.0.0-alpha1, 3.0.0-alpha2
>            Reporter: Reid Chan
>            Priority: Minor
>
> There's a case in my company: when two namenodes tried to transit state, checkpoint took
place in standbynode and it didn't check cancellation until every after 4096 inodes saved,
which took too much time between checks in our case and caused state transition fail at the
end. After we changed it to 50, state transition succeed.
> The *CHECK_CANCEL_INTERVAL* was introduced in HDFS-7097 and set to 4096, previously was
50. There may be some reasons for the choice of 4096, but i'm not sure that, so I suggest
to make it configurable.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org


Mime
View raw message