[JBoss JIRA] (WFLY-12929) Session remove event can deadlock if it attempts write operations on a session
by Paul Ferraro (Jira)
[ https://issues.redhat.com/browse/WFLY-12929?page=com.atlassian.jira.plugi... ]
Paul Ferraro updated WFLY-12929:
--------------------------------
Affects Version/s: 18.0.1.Final
(was: 18.0.0.Final)
> Session remove event can deadlock if it attempts write operations on a session
> ------------------------------------------------------------------------------
>
> Key: WFLY-12929
> URL: https://issues.redhat.com/browse/WFLY-12929
> Project: WildFly
> Issue Type: Bug
> Components: Clustering
> Affects Versions: 18.0.1.Final
> Reporter: Paul Ferraro
> Assignee: Paul Ferraro
> Priority: Critical
> Labels: downstream_dependency
>
> Activation/passivation listeners are intentionally non-transactional - and thus should never attempt to perform cache writes.
> In order to trigger the requisite activation/passivation listeners, activation/passivation events need to lookup the cache entries relevant to a given session via SessionFactory.findValue(..). However, if there are entries missing (e.g. a creation meta data entry w/out a access meta data entry), this method will attempt to purge the orphaned entries. This should never be done within the context of an activation/passivation event.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 5 months
[JBoss JIRA] (WFLY-12929) Session remove event can deadlock if it attempts write operations on a session
by Paul Ferraro (Jira)
[ https://issues.redhat.com/browse/WFLY-12929?page=com.atlassian.jira.plugi... ]
Paul Ferraro updated WFLY-12929:
--------------------------------
Fix Version/s: (was: 19.0.0.Beta1)
> Session remove event can deadlock if it attempts write operations on a session
> ------------------------------------------------------------------------------
>
> Key: WFLY-12929
> URL: https://issues.redhat.com/browse/WFLY-12929
> Project: WildFly
> Issue Type: Bug
> Components: Clustering
> Affects Versions: 18.0.1.Final
> Reporter: Paul Ferraro
> Assignee: Paul Ferraro
> Priority: Critical
> Labels: downstream_dependency
>
> Activation/passivation listeners are intentionally non-transactional - and thus should never attempt to perform cache writes.
> In order to trigger the requisite activation/passivation listeners, activation/passivation events need to lookup the cache entries relevant to a given session via SessionFactory.findValue(..). However, if there are entries missing (e.g. a creation meta data entry w/out a access meta data entry), this method will attempt to purge the orphaned entries. This should never be done within the context of an activation/passivation event.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 5 months
[JBoss JIRA] (WFLY-12929) Session remove event can deadlock if it attempts write operations on a session
by Paul Ferraro (Jira)
[ https://issues.redhat.com/browse/WFLY-12929?page=com.atlassian.jira.plugi... ]
Paul Ferraro updated WFLY-12929:
--------------------------------
Git Pull Request: (was: https://github.com/wildfly/wildfly/pull/12705)
> Session remove event can deadlock if it attempts write operations on a session
> ------------------------------------------------------------------------------
>
> Key: WFLY-12929
> URL: https://issues.redhat.com/browse/WFLY-12929
> Project: WildFly
> Issue Type: Bug
> Components: Clustering
> Affects Versions: 18.0.1.Final
> Reporter: Paul Ferraro
> Assignee: Paul Ferraro
> Priority: Critical
> Labels: downstream_dependency
>
> Activation/passivation listeners are intentionally non-transactional - and thus should never attempt to perform cache writes.
> In order to trigger the requisite activation/passivation listeners, activation/passivation events need to lookup the cache entries relevant to a given session via SessionFactory.findValue(..). However, if there are entries missing (e.g. a creation meta data entry w/out a access meta data entry), this method will attempt to purge the orphaned entries. This should never be done within the context of an activation/passivation event.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 5 months
[JBoss JIRA] (WFLY-12929) Session remove event can deadlock if it attempts write operations on a session
by Paul Ferraro (Jira)
Paul Ferraro created WFLY-12929:
-----------------------------------
Summary: Session remove event can deadlock if it attempts write operations on a session
Key: WFLY-12929
URL: https://issues.redhat.com/browse/WFLY-12929
Project: WildFly
Issue Type: Bug
Components: Clustering
Affects Versions: 18.0.0.Final
Reporter: Paul Ferraro
Assignee: Paul Ferraro
Fix For: 19.0.0.Beta1
Activation/passivation listeners are intentionally non-transactional - and thus should never attempt to perform cache writes.
In order to trigger the requisite activation/passivation listeners, activation/passivation events need to lookup the cache entries relevant to a given session via SessionFactory.findValue(..). However, if there are entries missing (e.g. a creation meta data entry w/out a access meta data entry), this method will attempt to purge the orphaned entries. This should never be done within the context of an activation/passivation event.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 5 months
[JBoss JIRA] (WFLY-12915) Clean out unneeded provisioning boilerplate
by Brian Stansberry (Jira)
[ https://issues.redhat.com/browse/WFLY-12915?page=com.atlassian.jira.plugi... ]
Brian Stansberry resolved WFLY-12915.
-------------------------------------
Resolution: Duplicate Issue
> Clean out unneeded provisioning boilerplate
> -------------------------------------------
>
> Key: WFLY-12915
> URL: https://issues.redhat.com/browse/WFLY-12915
> Project: WildFly
> Issue Type: Task
> Components: Build System
> Reporter: Brian Stansberry
> Assignee: Brian Stansberry
> Priority: Major
>
> The current Galleon versions should allow trimming quite a bit of boilerplate from all the places where we provision servers:
> Alexey Loubyansky 5:07 PM
> I did a bit more work on this, fixed a couple of issues and just released Galleon 4.2.1.Final and WFGP 4.2.1.Final
> Alexey Loubyansky 5:07 PM
> also opened a PR to upgrade WildFly Core to these releases https://github.com/wildfly/wildfly-core/pull/4025
> 5:08 PM
> upgrading doesn't affect anything, except the bug fixes contained in those versions
> 5:10 PM
> however, for wildfly, i think we should use GAVs as feature-pack locations for feature-pack dependencies, in place of Galleon FPL
> 5:11 PM
> the benefit would be to avoid the universe resolver in maven builds
> 5:11 PM
> and also a simplified (mojo) provisioning config
> Brian Stansberry 5:14 PM
> good timing; I'm working on how to add a new f-p to wildfly (for microprofile stuff) so i should use best practices :)
> Alexey Loubyansky 5:16 PM
> by simplified provisioning config i mean that the selected block should be removed completely https://github.com/wildfly/wildfly/blob/master/build/pom.xml#L84-L97
> 5:16 PM
> the only reason it is there is to avoid resolution from the universe
> 5:17 PM
> with GAVs for dependencies, it won't be necessary anymore
> Brian Stansberry 5:17 PM
> what about <inheritConfigs>false</inheritConfigs> ?
> Alexey Loubyansky 5:18 PM
> that was there to workaround a bug which was fixed in 4.2.0.Final
> Brian Stansberry 5:18 PM
> how about
> 5:18 PM
> <included-packages>
> <name>docs.examples.configs</name>
> </included-packages>
> 5:19 PM
> assuming that's not in the top level f-p
> Alexey Loubyansky 5:19 PM
> that should definitely stay, it's not in the selected block
> 5:19 PM
> ah, then you can have that
> 5:20 PM
> then there would be a real reason to mention the transitive dependency
> 5:20 PM
> i.e. you want something from it
> Brian Stansberry 5:20 PM
> could it be declared in the top level block, even though the package is in the transitive dep?
> 5:20 PM
> i'm going off on tangents, but first i should have said...
> 5:21 PM
> Nice!
> Alexey Loubyansky 5:21 PM (EDITED)
> unless the package is installed by default, it has to be included explicitly using transitive element
>
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 5 months
[JBoss JIRA] (DROOLS-4903) In constraints fail to compile if negated or inside a forall pattern
by Matteo Casalino (Jira)
Matteo Casalino created DROOLS-4903:
---------------------------------------
Summary: In constraints fail to compile if negated or inside a forall pattern
Key: DROOLS-4903
URL: https://issues.redhat.com/browse/DROOLS-4903
Project: Drools
Issue Type: Bug
Components: core engine
Affects Versions: 7.31.0.Final
Reporter: Matteo Casalino
Assignee: Mario Fusco
Attachments: in-constraint-negation.tgz
It appears that the KieBuilder fails to compile the following pattern:
{noformat}
Pojo(!(x in (1,2,3)))
{noformat}
returning the following error:
{noformat}
Error Messages:
Message [id=1, kieBase=defaultKieBase, level=ERROR, path=rules.drl, line=4, column=0
text=Unable to Analyse Expression !(x in (1,2,3)):
[Error: unexpected token: ,2]
[Near : {... !(x in (1,2,3)) ....}]
{noformat}
Furthermore, as of Drools 7.29.0.Final, the following form will also fail to compile:
{noformat}
forall(Pojo(y==42, x in (1,2,3)))
{noformat}
with the following error:
{noformat}
Error Messages:
Message [id=1, kieBase=defaultKieBase, level=ERROR, path=rules.drl, line=4, column=0
text=Unable to Analyse Expression !(y==42) || !(x in (1,2,3)):
[Error: unexpected token: ,2]
[Near : {... !(y==42) || !(x in (1,2,3)) ....}]
{noformat}
The latter is likely a consequence of the negation introduced by [DROOLS-4632|https://issues.redhat.com/browse/DROOLS-4632]
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 5 months