[infinispan-issues] [JBoss JIRA] (ISPN-1615) Consider removing transaction suspension code from putForExternalRead()

[Mircea Markus (Commented) (JIRA)]

    [ https://issues.jboss.org/browse/ISPN-1615?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12650952#comment-12650952 ] 

Mircea Markus commented on ISPN-1615:
-------------------------------------

As galder pointed out, the PFER does not guarnatee atomicity as it is async, so my prev comment doesn't stand.
Anyway below is a UT that reproduces another issue to be solved by this fix:
{code}
public class NoAutoCommitAndPferTest extends SingleCacheManagerTest {

   @Override
   protected EmbeddedCacheManager createCacheManager() throws Exception {
      Configuration dsc = getDefaultStandaloneConfig(true);
      dsc.fluent().transaction().autoCommit(false);
      return new DefaultCacheManager(dsc);
   }

   public void testPferNoAutoCommitExplicitTransaction() throws Exception {
      tm().begin();
      cache.putForExternalRead("k","v");
      tm().commit();
      assert cache.get("k").equals("v"); //here is the failure!
   }

   public void testPferNoAutoCommit() throws Exception {
      cache.putForExternalRead("k","v");
      assert cache.get("k").equals("v"); //here is the failure!
   }
}
{code}
                
> Consider removing transaction suspension code from putForExternalRead()
> -----------------------------------------------------------------------
>
>                 Key: ISPN-1615
>                 URL: https://issues.jboss.org/browse/ISPN-1615
>             Project: Infinispan
>          Issue Type: Enhancement
>    Affects Versions: 5.0.1.FINAL, 5.1.0.CR1
>            Reporter: Galder Zamarreño
>            Assignee: Galder Zamarreño
>             Fix For: 5.1.0.CR2, 5.1.0.FINAL
>
>
> The main reason for suspending the transaction was to avoid failures impacting the result of the transaction.
> However, ever since ISPN-1556 has been implemented, any FAIL_SILENT operation (like PFER) won't have an impact on the transaction if it fails, so this transaction suspension is not really needed.
> Removing the suspension code is apparently a performance booster in 2LC.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.jboss.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira