[JBoss JIRA] (ISPN-12364) If a node crashes leave the lock is not released in CONSISTENT reliablility
by Wolf-Dieter Fink (Jira)
[ https://issues.redhat.com/browse/ISPN-12364?page=com.atlassian.jira.plugi... ]
Wolf-Dieter Fink updated ISPN-12364:
------------------------------------
Steps to Reproduce:
* Start a 3 node cluster
* try a lock on each node
* stop one node
The behavior is expected to be the same with AVAILABLE and CONSISTENT reliability but it is different.
With CONSISTENT the event is only fired for some locks
With AVAILABLE the event is fired for all locks and the lock for the missing node is forced to release
reproducer code is here -> [https://github.com/wfink/infinispan.playground.lock]
ensure pom will use the correct version and the mainClass is 'org.infinispan.wfink.playground.lock.ClusteredLockEmbeddedConfig'
start 3 instance with exec:java and use the defined locks l1,l2,l3 each with a different node and stop one of them
repeat the test after changed resources/infinispan.xml reliablility attribute
was:
* Start a 3 node cluster
* try a lock on each node
* stop one node
The behavior is expected to be the same with AVAILABLE and CONSISTENT reliability but it is different.
With CONSISTENT the event is only fired for some locks
With AVAILABLE the event is fired for all locks and the lock for the missing node is forced to release
> If a node crashes leave the lock is not released in CONSISTENT reliablility
> ---------------------------------------------------------------------------
>
> Key: ISPN-12364
> URL: https://issues.redhat.com/browse/ISPN-12364
> Project: Infinispan
> Issue Type: Bug
> Components: Clustered Locks
> Affects Versions: 11.0.3.Final, 12.0.0.Dev04
> Reporter: Wolf-Dieter Fink
> Priority: Major
>
> If a node leave the lock held by this instance is unlocked with reliablilty AVAILABLE and all other members can use the lock.
> But with reliability CONSITENT the lock is still set and, as there is no longer an owner, the only way is to force the unlock from application code, but there is no knowledge that the lock does not have an alive owner
> This is not the expected behaviour.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months
[JBoss JIRA] (ISPN-12364) If a node crashes leave the lock is not released in CONSISTENT reliablility
by Wolf-Dieter Fink (Jira)
[ https://issues.redhat.com/browse/ISPN-12364?page=com.atlassian.jira.plugi... ]
Wolf-Dieter Fink updated ISPN-12364:
------------------------------------
Description:
If a node leave the lock held by this instance is unlocked with reliablilty AVAILABLE and all other members can use the lock.
But with reliability CONSITENT the lock is still set and, as there is no longer an owner, the only way is to force the unlock from application code, but there is no knowledge that the lock does not have an alive owner
This is not the expected behaviour.
was:
If a node leave graceful by shutting down the CacheManager or stop the process (CTRL-C) the lock held by this instance is unlocked and all other members can use the lock.
But if the process crash or is killed (UNIX kill -9) the lock is still set and, as there is no longer an owner, the only way is to force the unlock.
This is not the expected behaviour.
> If a node crashes leave the lock is not released in CONSISTENT reliablility
> ---------------------------------------------------------------------------
>
> Key: ISPN-12364
> URL: https://issues.redhat.com/browse/ISPN-12364
> Project: Infinispan
> Issue Type: Bug
> Components: Clustered Locks
> Affects Versions: 11.0.3.Final, 12.0.0.Dev04
> Reporter: Wolf-Dieter Fink
> Priority: Major
>
> If a node leave the lock held by this instance is unlocked with reliablilty AVAILABLE and all other members can use the lock.
> But with reliability CONSITENT the lock is still set and, as there is no longer an owner, the only way is to force the unlock from application code, but there is no knowledge that the lock does not have an alive owner
> This is not the expected behaviour.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months
[JBoss JIRA] (ISPN-12364) If a node crashes leave the lock is not released in CONSISTENT reliablility
by Wolf-Dieter Fink (Jira)
[ https://issues.redhat.com/browse/ISPN-12364?page=com.atlassian.jira.plugi... ]
Wolf-Dieter Fink updated ISPN-12364:
------------------------------------
Steps to Reproduce:
* Start a 3 node cluster
* try a lock on each node
* stop one node
The behavior is expected to be the same with AVAILABLE and CONSISTENT reliability but it is different.
With CONSISTENT the event is only fired for some locks
With AVAILABLE the event is fired for all locks and the lock for the missing node is forced to release
was:Start a 3 node cluster and kill a node graceful (working) or by use "kill -9" ungraceful to show the wrong behavior
> If a node crashes leave the lock is not released in CONSISTENT reliablility
> ---------------------------------------------------------------------------
>
> Key: ISPN-12364
> URL: https://issues.redhat.com/browse/ISPN-12364
> Project: Infinispan
> Issue Type: Bug
> Components: Clustered Locks
> Affects Versions: 11.0.3.Final, 12.0.0.Dev04
> Reporter: Wolf-Dieter Fink
> Priority: Major
>
> If a node leave graceful by shutting down the CacheManager or stop the process (CTRL-C) the lock held by this instance is unlocked and all other members can use the lock.
> But if the process crash or is killed (UNIX kill -9) the lock is still set and, as there is no longer an owner, the only way is to force the unlock.
> This is not the expected behaviour.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months
[JBoss JIRA] (ISPN-12364) If a node crashes leave the lock is not released in CONSISTENT reliablility
by Wolf-Dieter Fink (Jira)
[ https://issues.redhat.com/browse/ISPN-12364?page=com.atlassian.jira.plugi... ]
Wolf-Dieter Fink updated ISPN-12364:
------------------------------------
Summary: If a node crashes leave the lock is not released in CONSISTENT reliablility (was: If a node crashes and leave not graceful the lock is not released)
> If a node crashes leave the lock is not released in CONSISTENT reliablility
> ---------------------------------------------------------------------------
>
> Key: ISPN-12364
> URL: https://issues.redhat.com/browse/ISPN-12364
> Project: Infinispan
> Issue Type: Bug
> Components: Clustered Locks
> Affects Versions: 11.0.3.Final, 12.0.0.Dev04
> Reporter: Wolf-Dieter Fink
> Priority: Major
>
> If a node leave graceful by shutting down the CacheManager or stop the process (CTRL-C) the lock held by this instance is unlocked and all other members can use the lock.
> But if the process crash or is killed (UNIX kill -9) the lock is still set and, as there is no longer an owner, the only way is to force the unlock.
> This is not the expected behaviour.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months
[JBoss JIRA] (ISPN-12388) Integration with Knative Eventing
by Dan Berindei (Jira)
Dan Berindei created ISPN-12388:
-----------------------------------
Summary: Integration with Knative Eventing
Key: ISPN-12388
URL: https://issues.redhat.com/browse/ISPN-12388
Project: Infinispan
Issue Type: Feature Request
Components: Core
Reporter: Dan Berindei
Assignee: Dan Berindei
Fix For: 12.0.0.Final
Produce different types of cloud-events to integrate with KNative eventing:
* ISPN-11767 Emit Cloud Events for security audit events
* ISPN-11768 Emit Cloud Events for cache level notifications
* ISPN-11769 Emit Cloud Events for cache manager notifications
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months
[JBoss JIRA] (ISPN-12307) Improve media type parsing
by Gustavo Fernandes (Jira)
[ https://issues.redhat.com/browse/ISPN-12307?page=com.atlassian.jira.plugi... ]
Gustavo Fernandes updated ISPN-12307:
-------------------------------------
Status: Resolved (was: Pull Request Sent)
Fix Version/s: 12.0.0.Dev05
Resolution: Done
> Improve media type parsing
> --------------------------
>
> Key: ISPN-12307
> URL: https://issues.redhat.com/browse/ISPN-12307
> Project: Infinispan
> Issue Type: Enhancement
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Fix For: 12.0.0.Dev05
>
>
> Media type parsing has some problems with double-quoted parameters:
> * {{;}} and {{,}} are interpreted as parameter/media type separator even when they occur in a quoted parameter value
> * {{'}} is interpreted as a quoting character, the spec says only {{"}} must be used for quoting
> * Only spaces are allowed, not {{\t}}.
> * Spaces are allowed in too many places, e.g. {{type / subtype}} or {{parameter = value}}
> * Start and end quotes are removed from quoted parameter values, but the escaped characters inside are not unescaped.
> * Quotes are not added back in {{toString()}}, which makes deserialization fail if one of the parameters included an escaped double-quote (e.g. {{type/subtype; param="\""}}).
> Performance can also be improved, as we create many {{String}} instances via {{trim()}}, {{substring()}}, or {{split()}}.
> The clearest explanation of the media type grammar seems to be in
> https://tools.ietf.org/html/rfc7231#section-3.1.1.1
> https://tools.ietf.org/html/rfc7230#section-3.2.6
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months
[JBoss JIRA] (ISPN-12387) Merge infinispan-search-mapper module into infinispan-query module
by Gustavo Fernandes (Jira)
[ https://issues.redhat.com/browse/ISPN-12387?page=com.atlassian.jira.plugi... ]
Gustavo Fernandes updated ISPN-12387:
-------------------------------------
Status: Resolved (was: Pull Request Sent)
Resolution: Done
> Merge infinispan-search-mapper module into infinispan-query module
> ------------------------------------------------------------------
>
> Key: ISPN-12387
> URL: https://issues.redhat.com/browse/ISPN-12387
> Project: Infinispan
> Issue Type: Task
> Components: Embedded Querying
> Reporter: Nistor Adrian
> Assignee: Nistor Adrian
> Priority: Major
> Fix For: 12.0.0.Dev05
>
>
> This module existed just to isolate and ease HS6 integration effort but it has no purpose in itself as it is just a detail of query module implementation. All its code will be moved to query module. The HS backward compat annotation code will be removed later and external artifact will be used instead.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)
4 years, 12 months