[JBoss JIRA] (WFLY-10380) ThreadPoolExecutor leak under SM due to lack of privileged block
by David Lloyd (JIRA)
[ https://issues.jboss.org/browse/WFLY-10380?page=com.atlassian.jira.plugin... ]
David Lloyd updated WFLY-10380:
-------------------------------
Labels: security-manager (was: )
> ThreadPoolExecutor leak under SM due to lack of privileged block
> ----------------------------------------------------------------
>
> Key: WFLY-10380
> URL: https://issues.jboss.org/browse/WFLY-10380
> Project: WildFly
> Issue Type: Bug
> Components: JMS
> Reporter: David Lloyd
> Priority: Critical
> Labels: security-manager
>
> Still researching the source of these leaks.
> The way the leak happens is, a {{java.util.concurrent.ThreadPoolExecutor}} is constructed from an unprivileged context. The pool starts up and threads are created without a problem, however, the thread pool is never shut down. The finalizer runs but since it tries to shut down the pool with an access control context that was captured during construction, it fails because the context did not have the {{modifyThread}} {{RuntimePermission}}, and the thread pool never shuts down.
> We need to identify the points where TPEs are being constructed without controlled privileges.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
7 years, 7 months
[JBoss JIRA] (WFCORE-2086) Intermittent failure in OperationCancellationTestCase
by Richard Opalka (JIRA)
[ https://issues.jboss.org/browse/WFCORE-2086?page=com.atlassian.jira.plugi... ]
Richard Opalka edited comment on WFCORE-2086 at 5/31/18 12:30 PM:
------------------------------------------------------------------
I am also aware of this problem but didn't have time to analyze it. What I know about this issue is it was introduced with this commit:
---
[blackhole][/home/opalka/git/redhat/wildfly-core]>git log -1 remotes/origin/failing-10-runs-9-failures
commit 9bd9d19bbd39a06743b5434a8de14cd125d76294 (origin/failing-10-runs-9-failures)
Author: David M. Lloyd <david.lloyd(a)redhat.com>
Date: Thu Nov 2 13:38:14 2017 -0500
[WFCORE-3397] Introduce new thread pool into AbstractModelControllerOperationHandlerFactoryService
---
was (Author: ropalka):
I am also aware of this problem but didn't have time to analyze it. What I know about this issue is it was introduced with this commit:
---
[blackhole][/home/opalka/git/redhat/wildfly-core]>git log -1 remotes/origin/failing-10-runs-1-failures
commit 5145712087c9ae0625fc8c730e924967ecff9d83 (origin/failing-10-runs-1-failures)
Author: David M. Lloyd <david.lloyd(a)redhat.com>
Date: Thu Nov 2 08:51:12 2017 -0500
[WFCORE-3397] Introduce new thread pool to (deprecated) model controller client configuration factory
---
> Intermittent failure in OperationCancellationTestCase
> -----------------------------------------------------
>
> Key: WFCORE-2086
> URL: https://issues.jboss.org/browse/WFCORE-2086
> Project: WildFly Core
> Issue Type: Bug
> Components: Management
> Affects Versions: 5.0.0.Beta5
> Reporter: Brian Stansberry
> Assignee: Brian Stansberry
>
> Investigate this:
> https://ci.wildfly.org/viewLog.html?buildId=36108&buildTypeId=WildFlyCore...
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
7 years, 7 months
[JBoss JIRA] (ELY-1584) Add ability to log security manager violations to out-of-band log file
by David Lloyd (JIRA)
[ https://issues.jboss.org/browse/ELY-1584?page=com.atlassian.jira.plugin.s... ]
David Lloyd commented on ELY-1584:
----------------------------------
[~honza889] it might be sufficient. Mainly we'd want to see if it is possible to create a log of failures from the test suite (without missing anything, which is a challenge as each module may have a different logging configuration). And we'd want to ensure that the logged permission violations do not change unexpectedly from one test run to the next. Part of the challenge is that things do not always run in order though, so we might need a special comparison tool for this.
I was thinking that it might even be a good idea to have options for a specialized format which can be easily machine-parsed which includes the permission, the PD, and the stack trace.
> Add ability to log security manager violations to out-of-band log file
> ----------------------------------------------------------------------
>
> Key: ELY-1584
> URL: https://issues.jboss.org/browse/ELY-1584
> Project: WildFly Elytron
> Issue Type: Enhancement
> Components: Security Manager
> Reporter: David Lloyd
>
> It would be useful for CI if we could set a system property which would log security manager permission check violations (and the details thereof) to a log file, outside of the standard log infrastructure. This would potentially allow us to fail CI when unexpected "hidden" permission violations occur by way of a post-build script analyzes the log output.
> This might also be useful for troubleshooting more obscure security manager related problems.
--
This message was sent by Atlassian JIRA
(v7.5.0#75005)
7 years, 7 months