[
https://issues.apache.org/jira/browse/OOZIE-1398?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13694224#comment-13694224
]
Robert Kanter commented on OOZIE-1398:
--------------------------------------
I know I'm a bit late here, but I have a concern about the following change in
JsonWorkflowAction.java:
{code:java}
@@ -267,6 +265,9 @@
}
public void setData(String data) {
+ if( data != null && data.length() > 2000){
+ data = data.substring(0, 2000);
+ }
this.data = data;
}
{code}
This gets called to set the data when {{</capture-output>}} is used in an
action. Its max length can be set by the user in oozie-site via
{{oozie.action.max.output.data}} (or via
{{oozie.servlet.CallbackServlet.max.data.len}} because of OOZIE-4), which is
enforced before this method is called. The default value is 2K but if the user
sets it higher, won't this new code effectively hardcode it to 2K regardless of
what the user set? This means that the user will be losing data if they have
more than 2K.
Also, this String ({{data}}) was originally read in from a properties file, so
it should be a bunch of key-value pairs when we convert it to a String (I'm not
exactly sure what the format looks like though); if we simply truncate this
String, is it possible that we corrupt the last key-value pair if it doesn't
get cut off exactly between two key-value pairs (i.e. a key-value pair gets cut
somewhere in the middle)?
> [Scale] Reduce the number of CLOB columns used
> ----------------------------------------------
>
> Key: OOZIE-1398
> URL: https://issues.apache.org/jira/browse/OOZIE-1398
> Project: Oozie
> Issue Type: Improvement
> Affects Versions: trunk, 3.3.2
> Reporter: Rohini Palaniswamy
> Assignee: Ryota Egashira
> Fix For: trunk
>
> Attachments: OOZIE-1398-v4.patch, OOZIE-1398-v5.patch,
> OOZIE-1398-v7.patch, OOZIE-1398-v8.patch, OOZIE-1398-v8.patch
>
>
> When the number of concurrent submissions on Oozie increased to 100-200 per
> minute, it was not able to scale and we hit Oracle issues as there were lot
> of CLOB columns and DB became a bottle neck.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
