[infinispan-issues] [JBoss JIRA] (ISPN-11020) Clustered Max Idle Take 2

[Will Burns (Jira)]

    [ https://issues.redhat.com/browse/ISPN-11020?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13821497#comment-13821497 ] 

Will Burns commented on ISPN-11020:
-----------------------------------

Issues to work through:

# Expiration is currently done inside the DataContainer which doesn't have a non blocking based API (need to extract)
# Staggered gets will be problematic
# Stores and off heap need to be update values on retrieval
# Should putIfAbsent or other conditional commands update (putIfAbsent really should)

Nice to haves:
# Enhance to check for not having expired entries and bypassing for things like size?

> Clustered Max Idle Take 2
> -------------------------
>
>                 Key: ISPN-11020
>                 URL: https://issues.redhat.com/browse/ISPN-11020
>             Project: Infinispan
>          Issue Type: Feature Request
>          Components: Core, Expiration
>            Reporter: Will Burns
>            Assignee: Will Burns
>            Priority: Major
>
> ISPN-9003 added in clustered max idle expiration. However the approach implemented can cause entries to entries to expire prematurely when a node is lost (due to only the primary having the latest access time). This was deemed not acceptable.
> The new approach will add a new touch command that is fired on an entry access of a max idle entry. This will slow down accesses as they will essentially be light weight synchronous writes, but should guarantee data is properly expired.



--
This message was sent by Atlassian Jira
(v7.13.8#713008)