[JBoss JIRA] (WFLY-12881) Cannot customize split behavior and merge policy for Infinispan partition handling
by Paul Ferraro (Jira)
[ https://issues.redhat.com/browse/WFLY-12881?page=com.atlassian.jira.plugi... ]
Paul Ferraro updated WFLY-12881:
--------------------------------
Summary: Cannot customize split behavior and merge policy for Infinispan partition handling (was: Cannot customize split detection and merge policy for Infinispan partition handling)
> Cannot customize split behavior and merge policy for Infinispan partition handling
> ----------------------------------------------------------------------------------
>
> Key: WFLY-12881
> URL: https://issues.redhat.com/browse/WFLY-12881
> Project: WildFly
> Issue Type: Bug
> Components: Clustering
> Affects Versions: 18.0.1.Final
> Reporter: Paul Ferraro
> Assignee: Radoslav Husar
> Priority: Critical
> Fix For: 20.0.0.Beta1
>
>
> Currently, partition handling of an Infinispan cache is hard coded. When enabled, both reads and writes are denied on minority partitions (of a given segment) and, more critically, upon partition merge, no reconciliation of any data conflicts occurs.
> Users need to be able to configure this, at least to support the built in read/write on split policy and the built-in merge policies.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 3 months
[JBoss JIRA] (DROOLS-4919) Long running query at kie-server startup
by Cristiano Nicolai (Jira)
[ https://issues.redhat.com/browse/DROOLS-4919?page=com.atlassian.jira.plug... ]
Cristiano Nicolai commented on DROOLS-4919:
-------------------------------------------
[~swiderski.maciej] correct, that was done as part of JBPM-8559.
> Long running query at kie-server startup
> ----------------------------------------
>
> Key: DROOLS-4919
> URL: https://issues.redhat.com/browse/DROOLS-4919
> Project: Drools
> Issue Type: Bug
> Components: kie server
> Affects Versions: 7.28.0.Final
> Reporter: Danny Rucker
> Assignee: Maciej Swiderski
> Priority: Major
>
> The following query is executed at start up of kie-server
> ```
> select vil.processInstanceId, vil.processId, vil.id, vil.variableId, vil.value from VariableInstanceLog vil where vil.id in (select MAX(v.id) from VariableInstanceLog v group by v.variableId, v.processInstanceId)
> ```
> This is causing some issues with getting kie-server to start up.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 3 months
[JBoss JIRA] (JGRP-1860) Custom classloader in RpcDispatcher
by Todd Sandor (Jira)
[ https://issues.redhat.com/browse/JGRP-1860?page=com.atlassian.jira.plugin... ]
Todd Sandor commented on JGRP-1860:
-----------------------------------
We are a Redhat customer - this issue was raised on our behalf. We are not in the process of upgrading from JBoss EAP7.1 to 7.2 and the Groups version was changed from jgroups-3.6.14.Final-redhat-1.jar to jgroups-4.0.15.Final-redhat-00001.jar and at this point can not compile our application code (JGRP-1860).
These changes do not seem to have been ported to JGroups 4 (7.2.0 uses 4.0.15.Final-redhat-00001.jar). Is it possible to get these ported to JGroups 4?
I have raised case Redhat https://access.redhat.com/support/cases/#/case/02568685 for this.
We have also found that org.jgroups.util.UUID is not backwards compatible between the above releases ...but it is not related to JGRP-1860 .. The issues are described in the above case.
Cheers
> Custom classloader in RpcDispatcher
> -----------------------------------
>
> Key: JGRP-1860
> URL: https://issues.redhat.com/browse/JGRP-1860
> Project: JGroups
> Issue Type: Feature Request
> Affects Versions: 3.2.13
> Reporter: Dennis Reed
> Assignee: Bela Ban
> Priority: Major
> Fix For: 3.4.5, 3.5
>
>
> RpcDispatcher is hard-coded to use JGroups' classloader when marshalling the users's custom objects over RPC.
> RpcDispatcher uses Util.objectFromByteBuffer to unmarshall, which uses an ObjectInputStream. ObjectInputStream uses the classloader of its caller's class (Util).
> RpcDispatcher does allow a custom marshaller to be used (implementing RpcDispatcher.Marshaller), but since Util.objectFromByteBuffer hard-codes the use of ObjectInputStream, a custom marshaller cannot simply set the classloader and then delegate back to the default JGroups code.
> Util.objectFromBuffer should be enhanced to use a custom ObjectInputStream implementation that overrides resolveClass to use a custom classloader, and an API should be added to RpcDispatcher to pass in the classloader to use.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 3 months
[JBoss JIRA] (WFCORE-4781) Ability to enable shutdown operation for EmbeddedServer
by Jean Francois Denise (Jira)
[ https://issues.redhat.com/browse/WFCORE-4781?page=com.atlassian.jira.plug... ]
Jean Francois Denise commented on WFCORE-4781:
----------------------------------------------
An embedded server can't be shutdown. Would be needed by the maven-plugin to kill a server started in background. Would be needed in general to implement termination strategy using CLI.
> Ability to enable shutdown operation for EmbeddedServer
> -------------------------------------------------------
>
> Key: WFCORE-4781
> URL: https://issues.redhat.com/browse/WFCORE-4781
> Project: WildFly Core
> Issue Type: Enhancement
> Components: Embedded
> Affects Versions: 11.0.0.Beta5
> Reporter: Jean Francois Denise
> Assignee: James Perkins
> Priority: Major
>
> Having the shutdown operation available for embedded is needed when the embedder process and embedded server shutdown phase are bound.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 3 months
[JBoss JIRA] (WFCORE-4485) Support for multiple security realms - Distributed Identities
by Carl Walker (Jira)
[ https://issues.redhat.com/browse/WFCORE-4485?page=com.atlassian.jira.plug... ]
Carl Walker commented on WFCORE-4485:
-------------------------------------
I have two use cases. In both cases, authentication and authorization info is kept in multiple places. (This isn't the "authenticate here / authorize there" scenario supported with aggregate.)
1) End users and admins credentials are located in different datastores (tables) but access the same resources. End users hit the first query. If that passes, they're authenticated and authorized. Admins fail on the first query, but succeed on the second. Business logic keeps the usernames unique across both stores.
2) Customer-maintained Active Directory and RDBMS-oriented "system accounts". Authentication is done against Active Directory for end users who have an identity in the enterprise system. System accounts will fail the AD check but will be picked up in a second RDBMS module.
> Support for multiple security realms - Distributed Identities
> -------------------------------------------------------------
>
> Key: WFCORE-4485
> URL: https://issues.redhat.com/browse/WFCORE-4485
> Project: WildFly Core
> Issue Type: Feature Request
> Components: Security
> Reporter: Farah Juma
> Priority: Major
> Labels: CD17-Deferred, EAP-CD19, Previous_RFE
> Fix For: 11.0.0.Beta8
>
>
> By stacking LoginModules it was possible using PicketBox to attempt to authenticate using one remote store and if that failed try the next store in the list.
> This RFE is to consider the use case where identities could be located across multiple stores and how they are aggregated together.
> Additionally this use case should consider how the authorization information could be loaded from multiple sources and merged.
> This RFE is not about fail over in the event of a realm being unavailable although it may be related.
> This RFE is created as a result of comparing the differences between the PicketBox JAAS architecture and the Elytron architecture so I would not recommend this proceeds without some real world use cases identified.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 3 months