[JBoss JIRA] (ISPN-8822) Performance degradation when running Spark jobs
by Gustavo Fernandes (JIRA)
[ https://issues.jboss.org/browse/ISPN-8822?page=com.atlassian.jira.plugin.... ]
Gustavo Fernandes commented on ISPN-8822:
-----------------------------------------
Indeed, my proposal is to call shutdowngracefully() with an infinitesimal small of time and make close() not wait for any new potential new task. At least in Spark, there is no possibility of that happen, once I close the RCM no tasks will ever be aubmitted.
> Performance degradation when running Spark jobs
> -----------------------------------------------
>
> Key: ISPN-8822
> URL: https://issues.jboss.org/browse/ISPN-8822
> Project: Infinispan
> Issue Type: Bug
> Components: Hot Rod
> Affects Versions: 9.2.0.CR2
> Reporter: Gustavo Fernandes
>
> When trying to upgrade the [Spark connector |https://github.com/infinispan/infinispan-spark/] to use Infinispan 9.2.0.CR2, the test suite runs about 10x slower.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
8 years, 1 month
[JBoss JIRA] (ISPN-6471) Avoid blocking on Netty IO threads
by Radim Vansa (JIRA)
[ https://issues.jboss.org/browse/ISPN-6471?page=com.atlassian.jira.plugin.... ]
Radim Vansa closed ISPN-6471.
-----------------------------
Resolution: Duplicate Issue
> Avoid blocking on Netty IO threads
> ----------------------------------
>
> Key: ISPN-6471
> URL: https://issues.jboss.org/browse/ISPN-6471
> Project: Infinispan
> Issue Type: Enhancement
> Components: Server
> Reporter: Gustavo Fernandes
>
> Currently all cache operations are being executed in the {{NioEventLoopGroup}}.
> Threads from this event loop group are responsible to process IO events from one or more channels and if it's blocked by one cache operation, all channels will stop processing events.
> We should not block on IO threads but rather on a separate executor.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
8 years, 1 month
[JBoss JIRA] (ISPN-8822) Performance degradation when running Spark jobs
by Radim Vansa (JIRA)
[ https://issues.jboss.org/browse/ISPN-8822?page=com.atlassian.jira.plugin.... ]
Radim Vansa commented on ISPN-8822:
-----------------------------------
[~gustavonalle] The 2s delay is introduced in {{ChannelFactory.destroy}}: {{eventLoopGroup.shutdownGracefully().sync().getNow();}} uses 2 second delay:
{quote}
Graceful shutdown ensures that no tasks are submitted for 'the quiet period' (usually a couple seconds) before it shuts itself down. If a task is submitted during the quiet period, it is guaranteed to be accepted and the quiet period will start over.
{quote}
... and that method is preferred compared to regular {{shutdown}}. We could try to find out why Netty uses this pattern to closing the event loop.
> Performance degradation when running Spark jobs
> -----------------------------------------------
>
> Key: ISPN-8822
> URL: https://issues.jboss.org/browse/ISPN-8822
> Project: Infinispan
> Issue Type: Bug
> Components: Hot Rod
> Affects Versions: 9.2.0.CR2
> Reporter: Gustavo Fernandes
>
> When trying to upgrade the [Spark connector |https://github.com/infinispan/infinispan-spark/] to use Infinispan 9.2.0.CR2, the test suite runs about 10x slower.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
8 years, 1 month
[JBoss JIRA] (ISPN-8838) Upgrade to Hibernate ORM 5.2.12.Final
by Sanne Grinovero (JIRA)
Sanne Grinovero created ISPN-8838:
-------------------------------------
Summary: Upgrade to Hibernate ORM 5.2.12.Final
Key: ISPN-8838
URL: https://issues.jboss.org/browse/ISPN-8838
Project: Infinispan
Issue Type: Task
Components: Test Suite - Query
Reporter: Sanne Grinovero
Assignee: Sanne Grinovero
Fix For: 9.2.0.Final
This upgrade is required to complete the upgrade of Hibernate Search ISPN-8779
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
8 years, 1 month