[JBoss JIRA] (WFLY-13269) The org.wildfly:wildfly-client-all artefact incorrectly contains the Elytron CDI extension
by Jeff Mesnil (Jira)
[ https://issues.redhat.com/browse/WFLY-13269?page=com.atlassian.jira.plugi... ]
Jeff Mesnil commented on WFLY-13269:
------------------------------------
[~jamezp] yes, I've added it.
> The org.wildfly:wildfly-client-all artefact incorrectly contains the Elytron CDI extension
> ------------------------------------------------------------------------------------------
>
> Key: WFLY-13269
> URL: https://issues.redhat.com/browse/WFLY-13269
> Project: WildFly
> Issue Type: Bug
> Components: Build System, MP JWT, Security
> Reporter: Darran Lofthouse
> Assignee: Darran Lofthouse
> Priority: Critical
> Fix For: 20.0.0.Beta1
>
>
> This means if the artefact is included in a deployment the JWT activation is incorrectly triggered but with some dependencies missing leading to an error such as the following.
> {code}
> 09:46:44,391 WARN [org.jboss.modules.define] (MSC service thread 1-1) Failed to define class org.wildfly.security.mp.jwt.JWTCDIExtension in Module "deployment.XXXX.ear" from Service Module Loader: java.lang.NoClassDefFoundError: Failed to link org/wildfly/security/mp/jwt/JWTCDIExtension (Module "deployment.XXXX.ear" from Service Module Loader): io/smallrye/jwt/auth/cdi/SmallRyeJWTAuthCDIExtension
> at java.lang.ClassLoader.defineClass1(ClassLoader.java)
> {code}
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 1 month
[JBoss JIRA] (JGRP-2462) TransferQueueBundler: drop messages if queue is full
by Bela Ban (Jira)
[ https://issues.redhat.com/browse/JGRP-2462?page=com.atlassian.jira.plugin... ]
Bela Ban edited comment on JGRP-2462 at 3/25/20 12:28 PM:
----------------------------------------------------------
Protocols such as RED or Blue (\[2\]) have to know the queue's capacity and the current size, e.g. to maintain an average size, which is used to compute the probability with which a message is dropped. The capacity is fixed and can be fetched via {{TP.getBundlerCapacity()}}.
Seems like adding {{getQueueSize()}} to {{Bundler}} would allow us to implement random early detection in a separate protocol (e.g. {{RED}}). This is *not* the same as {{Bundler.size()}}; we cannot use the latter as it returns the total number of _unsent_ messages.
\[2\\ https://en.wikipedia.org/wiki/Blue_(queue_management_algorithm)
was (Author: belaban):
Protocols such as RED or Blue (\[2\]) have to know the queue's capacity and the current size, e.g. to maintain an average size, which is used to compute the probability with which a message is dropped.
Seems like adding {{getQueueSize()}} to {{Bundler}} would allow us to implement random early detection in a separate protocol (e.g. {RED}}). This is *not* the same as {{Bundler.size()}}; we cannot use the latter as it returns the total number of _unsent_ messages.
\[2\\ https://en.wikipedia.org/wiki/Blue_(queue_management_algorithm)
> TransferQueueBundler: drop messages if queue is full
> ----------------------------------------------------
>
> Key: JGRP-2462
> URL: https://issues.redhat.com/browse/JGRP-2462
> Project: JGroups
> Issue Type: Feature Request
> Reporter: Bela Ban
> Assignee: Bela Ban
> Priority: Major
> Fix For: 4.2.2, 5.0.0.Alpha4
>
>
> When a bundler (like {{TransferQueueBundler}}) has a full queue, the senders will block.
> This can happen for example when the TCP {{send}} blocks due to a full send-window.
> Retransmission requests and responses compound the problem when the TQB's queue is full.
> Therefore, in addition to blocking, we should add an option to _drop_ packets when the queue is full. Investigate whether we could use some priority based scheme, which drops non-important messages first.
> Also investigate whether senders can selectively be unblocked when the queue is full, to either drop or add their messages. This could be done for example by removing elements from the queue, using the aformentioned filter.
> Look at random early detection )\[1\]) for inspiration.
> \[1\] https://en.wikipedia.org/wiki/Random_early_detection
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 1 month
[JBoss JIRA] (JGRP-2462) TransferQueueBundler: drop messages if queue is full
by Bela Ban (Jira)
[ https://issues.redhat.com/browse/JGRP-2462?page=com.atlassian.jira.plugin... ]
Bela Ban edited comment on JGRP-2462 at 3/25/20 12:28 PM:
----------------------------------------------------------
Protocols such as RED or Blue (\[2\]) have to know the queue's capacity and the current size, e.g. to maintain an average size, which is used to compute the probability with which a message is dropped. The capacity is fixed and can be fetched via {{TP.getBundlerCapacity()}}.
Seems like adding {{getQueueSize()}} to {{Bundler}} would allow us to implement random early detection in a separate protocol (e.g. {{RED}}). This is *not* the same as {{Bundler.size()}}; we cannot use the latter as it returns the total number of _unsent_ messages.
\[2\] https://en.wikipedia.org/wiki/Blue_(queue_management_algorithm)
was (Author: belaban):
Protocols such as RED or Blue (\[2\]) have to know the queue's capacity and the current size, e.g. to maintain an average size, which is used to compute the probability with which a message is dropped. The capacity is fixed and can be fetched via {{TP.getBundlerCapacity()}}.
Seems like adding {{getQueueSize()}} to {{Bundler}} would allow us to implement random early detection in a separate protocol (e.g. {{RED}}). This is *not* the same as {{Bundler.size()}}; we cannot use the latter as it returns the total number of _unsent_ messages.
\[2\\ https://en.wikipedia.org/wiki/Blue_(queue_management_algorithm)
> TransferQueueBundler: drop messages if queue is full
> ----------------------------------------------------
>
> Key: JGRP-2462
> URL: https://issues.redhat.com/browse/JGRP-2462
> Project: JGroups
> Issue Type: Feature Request
> Reporter: Bela Ban
> Assignee: Bela Ban
> Priority: Major
> Fix For: 4.2.2, 5.0.0.Alpha4
>
>
> When a bundler (like {{TransferQueueBundler}}) has a full queue, the senders will block.
> This can happen for example when the TCP {{send}} blocks due to a full send-window.
> Retransmission requests and responses compound the problem when the TQB's queue is full.
> Therefore, in addition to blocking, we should add an option to _drop_ packets when the queue is full. Investigate whether we could use some priority based scheme, which drops non-important messages first.
> Also investigate whether senders can selectively be unblocked when the queue is full, to either drop or add their messages. This could be done for example by removing elements from the queue, using the aformentioned filter.
> Look at random early detection )\[1\]) for inspiration.
> \[1\] https://en.wikipedia.org/wiki/Random_early_detection
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 1 month
[JBoss JIRA] (JGRP-2462) TransferQueueBundler: drop messages if queue is full
by Bela Ban (Jira)
[ https://issues.redhat.com/browse/JGRP-2462?page=com.atlassian.jira.plugin... ]
Bela Ban commented on JGRP-2462:
--------------------------------
Protocols such as RED or Blue (\[2\]) have to know the queue's capacity and the current size, e.g. to maintain an average size, which is used to compute the probability with which a message is dropped.
Seems like adding {{getQueueSize()}} to {{Bundler}} would allow us to implement random early detection in a separate protocol (e.g. {RED}}). This is *not* the same as {{Bundler.size()}}; we cannot use the latter as it returns the total number of _unsent_ messages.
\[2\\ https://en.wikipedia.org/wiki/Blue_(queue_management_algorithm)
> TransferQueueBundler: drop messages if queue is full
> ----------------------------------------------------
>
> Key: JGRP-2462
> URL: https://issues.redhat.com/browse/JGRP-2462
> Project: JGroups
> Issue Type: Feature Request
> Reporter: Bela Ban
> Assignee: Bela Ban
> Priority: Major
> Fix For: 4.2.2, 5.0.0.Alpha4
>
>
> When a bundler (like {{TransferQueueBundler}}) has a full queue, the senders will block.
> This can happen for example when the TCP {{send}} blocks due to a full send-window.
> Retransmission requests and responses compound the problem when the TQB's queue is full.
> Therefore, in addition to blocking, we should add an option to _drop_ packets when the queue is full. Investigate whether we could use some priority based scheme, which drops non-important messages first.
> Also investigate whether senders can selectively be unblocked when the queue is full, to either drop or add their messages. This could be done for example by removing elements from the queue, using the aformentioned filter.
> Look at random early detection )\[1\]) for inspiration.
> \[1\] https://en.wikipedia.org/wiki/Random_early_detection
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 1 month