[JBoss JIRA] (WFCORE-4837) (7.3.z) [WFCORE-4596] Write lock is acquired reading patching resource using include-runtime
by Bartosz Spyrko (Jira)
Bartosz Spyrko created WFCORE-4837:
--------------------------------------
Summary: (7.3.z) [WFCORE-4596] Write lock is acquired reading patching resource using include-runtime
Key: WFCORE-4837
URL: https://issues.redhat.com/browse/WFCORE-4837
Project: WildFly Core
Issue Type: Bug
Components: Patching
Reporter: Bartosz Spyrko
Assignee: Yeray Borges
Fix For: 11.0.0.Beta3
This is initially created as a bug but it needs first some investigation to discover if it is really necessary to acquire the write lock when the patching resource is read. If it is unnecessary, we should remove this need.
The following operation is acquiring the write lock:
{noformat}
/host=slave/core-service=patching:read-resource(include-runtime=true)
{noformat}
It potentially will block if there is any other operation that has already acquired the lock. One consequence found due to this issue was HAL could block and hang if an HC is starting and the user clicks on 'Patching' menu entry. It was fixed in HAL, where an HC being restarted is no longer shown in the Patching menu.
The main problem seems to be in PatchStreamResourceOperationStepHandler, which is the parent handler used for the handlers that manage patching operations. This parent handler is always acquiring the write lock.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 5 months
[JBoss JIRA] (WFCORE-4836) (7.3.z) Validate composite operation steps just before executing them
by Bartosz Spyrko (Jira)
Bartosz Spyrko created WFCORE-4836:
--------------------------------------
Summary: (7.3.z) Validate composite operation steps just before executing them
Key: WFCORE-4836
URL: https://issues.redhat.com/browse/WFCORE-4836
Project: WildFly Core
Issue Type: Enhancement
Components: Management
Reporter: Bartosz Spyrko
Assignee: Yeray Borges
Fix For: 9.0.0.Beta7, 9.0.0.Final
Say we have a composite operation with 2 steps:
1) /extension=org.jboss.as.messaging:add
2) /subsystem=messaging:add
This will fail:
Failed to execute batch: JBAS014739: No handler for add at address
[("subsystem" => "messaging")]
This fails because at the time of validation the /subsystem=messaging:add is not valid.
To illustrate, the execution order is
Validate 1-2
1
2
A possible solution is to convert this to the following:
V1
1
V2 (works now because 1 has registered the subsystem API)
2
I think that should work but it's a very complex area, particularly in a managed domain, so it's not at all certain this would prove feasible.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
6 years, 5 months