[JBoss JIRA] (ISPN-9982) Test deadlock for clustered caches with Expiration enabled
by William Burns (Jira)
[ https://issues.jboss.org/browse/ISPN-9982?page=com.atlassian.jira.plugin.... ]
William Burns updated ISPN-9982:
--------------------------------
Summary: Test deadlock for clustered caches with Expiration enabled (was: Possible deadlock for clustered caches with Expiration enabled)
> Test deadlock for clustered caches with Expiration enabled
> ----------------------------------------------------------
>
> Key: ISPN-9982
> URL: https://issues.jboss.org/browse/ISPN-9982
> Project: Infinispan
> Issue Type: Bug
> Reporter: Diego Lovison
> Assignee: Diego Lovison
> Priority: Minor
> Fix For: 10.0.0.Beta3
>
>
> Add a test case for the following scenario:
> If a cache is configured for expiration an the intervall is != -1 there is a possiblity for a deadlock with JGroups and Infinispan threads because the reaper and any access to the same entry can cause a deadlock.
> The possibility is higher if the configured interval for the reaper is shorter.
> A thread dump might show something like followed:
> "HotRod-hotrod-internalServerWorker-4-12" #279 prio=5 os_prio=0 tid=0x00007f35980b8800 nid=0xbbb waiting for monitor entry [0x00007f356cfb4000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8.compute(EquivalentConcurrentHashMapV8.java:1910)
> - waiting to lock <0x0000000742cca028> (a org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8$Node)
> at org.infinispan.container.DefaultDataContainer.compute(DefaultDataContainer.java:335)
> at org.infinispan.expiration.impl.ExpirationManagerImpl.handleInMemoryExpiration(ExpirationManagerImpl.java:135)
> at org.infinispan.expiration.impl.ClusterExpirationManager.handleInMemoryExpiration(ClusterExpirationManager.java:152)
> - locked <0x0000000742cc9898> (a org.infinispan.container.entries.metadata.MetadataTransientCacheEntry)
> at org.infinispan.container.DefaultDataContainer.get(DefaultDataContainer.java:201)
> "pool-7-thread-1" #158 prio=5 os_prio=0 tid=0x00007f35c816c000 nid=0xb35 waiting for monitor entry [0x00007f3575238000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at org.infinispan.expiration.impl.ExpirationManagerImpl.lambda$handleInMemoryExpiration$0(ExpirationManagerImpl.java:137)
> - waiting to lock <0x0000000742cc9898> (a org.infinispan.container.entries.metadata.MetadataTransientCacheEntry)
> at org.infinispan.expiration.impl.ExpirationManagerImpl$$Lambda$374/1904915245.compute(Unknown Source)
> at org.infinispan.container.DefaultDataContainer.lambda$compute$3(DefaultDataContainer.java:336)
> at org.infinispan.container.DefaultDataContainer$$Lambda$375/1917382785.apply(Unknown Source)
> at org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8.compute(EquivalentConcurrentHashMapV8.java:1919)
> - locked <0x0000000742cca028> (a org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8$Node)
> at org.infinispan.container.DefaultDataContainer.compute(DefaultDataContainer.java:335)
> at org.infinispan.expiration.impl.ExpirationManagerImpl.handleInMemoryExpiration(ExpirationManagerImpl.java:135)
> at org.infinispan.expiration.impl.ClusterExpirationManager.processExpiration(ClusterExpirationManager.java:82)
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 8 months
[JBoss JIRA] (ISPN-9982) Possible deadlock for clustered caches with Expiration enabled
by William Burns (Jira)
[ https://issues.jboss.org/browse/ISPN-9982?page=com.atlassian.jira.plugin.... ]
William Burns commented on ISPN-9982:
-------------------------------------
To clarify this issue was fixed in ISPN-9003 for master.
> Possible deadlock for clustered caches with Expiration enabled
> --------------------------------------------------------------
>
> Key: ISPN-9982
> URL: https://issues.jboss.org/browse/ISPN-9982
> Project: Infinispan
> Issue Type: Bug
> Reporter: Diego Lovison
> Assignee: Diego Lovison
> Priority: Minor
> Fix For: 10.0.0.Beta3
>
>
> Add a test case for the following scenario:
> If a cache is configured for expiration an the intervall is != -1 there is a possiblity for a deadlock with JGroups and Infinispan threads because the reaper and any access to the same entry can cause a deadlock.
> The possibility is higher if the configured interval for the reaper is shorter.
> A thread dump might show something like followed:
> "HotRod-hotrod-internalServerWorker-4-12" #279 prio=5 os_prio=0 tid=0x00007f35980b8800 nid=0xbbb waiting for monitor entry [0x00007f356cfb4000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8.compute(EquivalentConcurrentHashMapV8.java:1910)
> - waiting to lock <0x0000000742cca028> (a org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8$Node)
> at org.infinispan.container.DefaultDataContainer.compute(DefaultDataContainer.java:335)
> at org.infinispan.expiration.impl.ExpirationManagerImpl.handleInMemoryExpiration(ExpirationManagerImpl.java:135)
> at org.infinispan.expiration.impl.ClusterExpirationManager.handleInMemoryExpiration(ClusterExpirationManager.java:152)
> - locked <0x0000000742cc9898> (a org.infinispan.container.entries.metadata.MetadataTransientCacheEntry)
> at org.infinispan.container.DefaultDataContainer.get(DefaultDataContainer.java:201)
> "pool-7-thread-1" #158 prio=5 os_prio=0 tid=0x00007f35c816c000 nid=0xb35 waiting for monitor entry [0x00007f3575238000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at org.infinispan.expiration.impl.ExpirationManagerImpl.lambda$handleInMemoryExpiration$0(ExpirationManagerImpl.java:137)
> - waiting to lock <0x0000000742cc9898> (a org.infinispan.container.entries.metadata.MetadataTransientCacheEntry)
> at org.infinispan.expiration.impl.ExpirationManagerImpl$$Lambda$374/1904915245.compute(Unknown Source)
> at org.infinispan.container.DefaultDataContainer.lambda$compute$3(DefaultDataContainer.java:336)
> at org.infinispan.container.DefaultDataContainer$$Lambda$375/1917382785.apply(Unknown Source)
> at org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8.compute(EquivalentConcurrentHashMapV8.java:1919)
> - locked <0x0000000742cca028> (a org.infinispan.commons.util.concurrent.jdk8backported.EquivalentConcurrentHashMapV8$Node)
> at org.infinispan.container.DefaultDataContainer.compute(DefaultDataContainer.java:335)
> at org.infinispan.expiration.impl.ExpirationManagerImpl.handleInMemoryExpiration(ExpirationManagerImpl.java:135)
> at org.infinispan.expiration.impl.ClusterExpirationManager.processExpiration(ClusterExpirationManager.java:82)
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 8 months
[JBoss JIRA] (ISPN-10022) Tx compute command updates are not written to stores
by Dan Berindei (Jira)
[ https://issues.jboss.org/browse/ISPN-10022?page=com.atlassian.jira.plugin... ]
Dan Berindei updated ISPN-10022:
--------------------------------
Status: Open (was: New)
> Tx compute command updates are not written to stores
> ----------------------------------------------------
>
> Key: ISPN-10022
> URL: https://issues.jboss.org/browse/ISPN-10022
> Project: Infinispan
> Issue Type: Bug
> Components: Core
> Affects Versions: 9.4.8.Final, 10.0.0.Beta2
> Reporter: Dan Berindei
> Priority: Major
> Fix For: 10.0.0.Beta3, 9.4.9.Final
>
>
> {{TxBatchUpdater}} collects the modifications from all the commands in a transaction in order to write them to the stores. However, it doesn't implement the {{visitCompute*}} commands, so it doesn't collect any modifications from {{compute()}}, {{computeIfAbsent()}}, and {{computeIfPresent()}}.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 8 months
[JBoss JIRA] (ISPN-10022) Tx compute command updates are not written to stores
by Dan Berindei (Jira)
[ https://issues.jboss.org/browse/ISPN-10022?page=com.atlassian.jira.plugin... ]
Dan Berindei reassigned ISPN-10022:
-----------------------------------
Assignee: Dan Berindei
> Tx compute command updates are not written to stores
> ----------------------------------------------------
>
> Key: ISPN-10022
> URL: https://issues.jboss.org/browse/ISPN-10022
> Project: Infinispan
> Issue Type: Bug
> Components: Core
> Affects Versions: 9.4.8.Final, 10.0.0.Beta2
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Fix For: 10.0.0.Beta3, 9.4.9.Final
>
>
> {{TxBatchUpdater}} collects the modifications from all the commands in a transaction in order to write them to the stores. However, it doesn't implement the {{visitCompute*}} commands, so it doesn't collect any modifications from {{compute()}}, {{computeIfAbsent()}}, and {{computeIfPresent()}}.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 8 months