[
https://issues.apache.org/jira/browse/IGNITE-770?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ivan Veselovsky updated IGNITE-770:
-----------------------------------
Description:
Current code of this method is:
{code}
@Override public void close() throws IOException {
if (cacheEnabled && get(getUri(), getConf()) == this)
return;
close0();
}
{code}
This code sometimes creates a new instance of the IgniteHadoopFileSystem in the
#get(2) call. This creation is expensive. Looks like this is not quite the
behavior what was expected by the author of code.
Need to investigate why that happens and possibly modify the code accordingly.
was:
Current code of this method is:
{code}
@Override public void close() throws IOException {
if (cacheEnabled && get(getUri(), getConf()) == this)
return;
close0();
}
{code}
This code creates a new instance of the IgniteHadoopFileSystem in #get(2) call.
This creation is quite expensive. Looks like this is not quite the behavior
what was expected by the author of code. Need to investigate why that happens
and possibly modify the code accordingly.
> Revise IgniteHadoopFileSystem#close() implementation
> -----------------------------------------------------
>
> Key: IGNITE-770
> URL: https://issues.apache.org/jira/browse/IGNITE-770
> Project: Ignite
> Issue Type: Bug
> Components: hadoop
> Affects Versions: sprint-4
> Reporter: Ivan Veselovsky
> Assignee: Ivan Veselovsky
>
> Current code of this method is:
> {code}
> @Override public void close() throws IOException {
> if (cacheEnabled && get(getUri(), getConf()) == this)
> return;
> close0();
> }
> {code}
> This code sometimes creates a new instance of the IgniteHadoopFileSystem in
> the #get(2) call. This creation is expensive. Looks like this is not quite
> the behavior what was expected by the author of code.
> Need to investigate why that happens and possibly modify the code accordingly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
