[
https://issues.apache.org/jira/browse/HBASE-19680?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chia-Ping Tsai updated HBASE-19680:
-----------------------------------
Status: Open (was: Patch Available)
Oh...run the patch with a bunch of threads. Ya, there is a concurrent issue...
The change to fix the concurrent issue is shown below.
{code:java}
AsyncRequestFuture asf;
try (QueueRowAccess access = new QueueRowAccess()) {
if (access.isEmpty()) {
// It means someone has gotten the ticker to run the flush.
break;
}
asf = ap.submit(createTask(access));
}
// DON'T do the wait in the try-with-resources. Otherwise, the undealt
mutations won't
// be released.
asf.waitUntilDone();
if (asf.hasError()) {
errors.add(asf.getErrors());
}{code}
> BufferedMutatorImpl#mutate should wait the result from AP in order to throw
> the failed mutations
> ------------------------------------------------------------------------------------------------
>
> Key: HBASE-19680
> URL: https://issues.apache.org/jira/browse/HBASE-19680
> Project: HBase
> Issue Type: Improvement
> Reporter: Chia-Ping Tsai
> Assignee: Chia-Ping Tsai
> Priority: Major
> Fix For: 2.0.0-beta-2
>
> Attachments: HBASE-19680.v0.patch, HBASE-19680.v1.patch
>
>
> Currently, BMI#mutate doesn't wait the result from AP so the errors are
> stored in AP. The only way which can return the errors to user is, calling
> the flush to catch the exception. That is non-intuitive.
> I feel BMI#mutate should wait the result. That is to say, user can parse the
> exception thrown by BM#mutate to get the failed mutations. Also, we can
> remove the global error from AP.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
