[JBoss JIRA] (ISPN-5575) Shared write-behind store can read stale entries on joiner
by William Burns (JIRA)
[ https://issues.jboss.org/browse/ISPN-5575?page=com.atlassian.jira.plugin.... ]
William Burns updated ISPN-5575:
--------------------------------
Fix Version/s: 9.0.0.Beta1
(was: 9.0.0.Alpha4)
> Shared write-behind store can read stale entries on joiner
> ----------------------------------------------------------
>
> Key: ISPN-5575
> URL: https://issues.jboss.org/browse/ISPN-5575
> Project: Infinispan
> Issue Type: Bug
> Components: Core, Loaders and Stores
> Affects Versions: 8.0.0.Alpha2, 7.2.3.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Fix For: 9.0.0.Beta1
>
>
> The AsyncCacheWriter modification queue is not sent with state transfer when the store is shared. A joiner can then read from the shared store a stale version of entries that have updates in the modification queue but are no longer in memory (because they were either removed explicitly, or evicted).
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)
7 years, 11 months
[JBoss JIRA] (ISPN-5574) Define high-level cache capabilities
by William Burns (JIRA)
[ https://issues.jboss.org/browse/ISPN-5574?page=com.atlassian.jira.plugin.... ]
William Burns updated ISPN-5574:
--------------------------------
Fix Version/s: 9.0.0.Beta1
(was: 9.0.0.Alpha4)
> Define high-level cache capabilities
> ------------------------------------
>
> Key: ISPN-5574
> URL: https://issues.jboss.org/browse/ISPN-5574
> Project: Infinispan
> Issue Type: Feature Request
> Components: Configuration, Core
> Affects Versions: 7.2.3.Final
> Reporter: Dan Berindei
> Priority: Minor
> Fix For: 9.0.0.Beta1
>
>
> Infinispan's configuration is very flexible, and it's sometimes hard to figure out how different settings affect things like cache consistency.
> For example, the lucene-directory module uses the fairly complicated {{Configurations.noDataLossOnJoiner()}} method to validate that a cache is safe for storing lucene indexes.
> Another example is users who would like to use a store for backup, but they don't want read from the store for M/R tasks or when get(k) doesn't find the key in memory.
> One idea would be to define a set of "capabilities" like "state-transfer-complete" or "all-data-in-memory". The user could then add those capabilities in the cache definition, and the cache won't start if the configuration violates those capabilities. The capabilities would also be used internally, to improve the error message when a feature requires a particular combination of settings.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)
7 years, 11 months
[JBoss JIRA] (ISPN-5614) Write performance regression after ISPN-5484
by William Burns (JIRA)
[ https://issues.jboss.org/browse/ISPN-5614?page=com.atlassian.jira.plugin.... ]
William Burns updated ISPN-5614:
--------------------------------
Fix Version/s: 9.0.0.Beta1
(was: 9.0.0.Alpha4)
> Write performance regression after ISPN-5484
> --------------------------------------------
>
> Key: ISPN-5614
> URL: https://issues.jboss.org/browse/ISPN-5614
> Project: Infinispan
> Issue Type: Bug
> Components: Core
> Affects Versions: 8.0.0.Beta1
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Fix For: 9.0.0.Beta1
>
>
> Regression test shows a significant drop in throughput in the replicated and distributed write tests.
> This was after adjusting the internal thread pool settings in the JGroups configuration: with the default (min=5, max=20, queue=0), the distributed read test would fail to finish.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)
7 years, 11 months
[JBoss JIRA] (ISPN-5642) Reduce contention in the RPC timeout handling
by William Burns (JIRA)
[ https://issues.jboss.org/browse/ISPN-5642?page=com.atlassian.jira.plugin.... ]
William Burns updated ISPN-5642:
--------------------------------
Fix Version/s: 9.0.0.Beta1
(was: 9.0.0.Alpha4)
> Reduce contention in the RPC timeout handling
> ---------------------------------------------
>
> Key: ISPN-5642
> URL: https://issues.jboss.org/browse/ISPN-5642
> Project: Infinispan
> Issue Type: Task
> Components: Core
> Affects Versions: 8.0.0.Beta2
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Minor
> Labels: help_wanted
> Fix For: 9.0.0.Beta1
>
>
> Most of the RPC timeout tasks are cancelled way before their timeout expires. This means the scheduler spends a lot of time reordering the elements of its DelayQueue.
> It should be possible to store the tasks with a long timeout (e.g. 1s) in a queue and only move them to the scheduler's priority queue when they have less than 1s to expiration (e.g. by a worker thread that runs every 0.5s)
> Storing all the tasks in a single queue may be impractical because the worker thread would have more and work to do as load increases and the RPCs take longer to finish, so a [hashed timing wheel|http://www.cse.wustl.edu/~cdgill/courses/cs6874/TimingWheels.ppt] may be needed.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)
7 years, 11 months