[infinispan-issues] [JBoss JIRA] (ISPN-2495) Replication on startup not reliable

[Thomas Fromm (JIRA)]

     [ https://issues.jboss.org/browse/ISPN-2495?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Thomas Fromm updated ISPN-2495:
-------------------------------

    Description: 
Situation: REPL cache with Cache Loader and ~500k entries and preload = true, and shared cache store.

Problem: 1st node starts and the startup (until cache loaded all entries and so on) takes lets say 2min. When starting the 2nd node after e.g. 30s, then state is requested, but contains no results. cache loader is also not used, since its not the 1st node. Log from 2nd node is attached.
So I've REPL cache over two (or more ) nodes, where only the 1st node contains the entries.

When starting all other nodes later, then state transfer works and the data gets replicated.

Maybe startup of all other nodes should be blocked until warm start of first node has finished?

  was:
Situation: REPL cache with Cache Loader and ~500k entries and preload = true

Problem: 1st node starts and the startup (until cache loaded all entries and so on) takes lets say 2min. When starting the 2nd node after e.g. 30s, then state is requested, but contains no results. cache loader is also not used, since its not the 1st node. Log from 2nd node is attached.
So I've REPL cache over two (or more ) nodes, where only the 1st node contains the entries.

When starting all other nodes later, then state transfer works and the data gets replicated.

Maybe startup of all other nodes should be blocked until warm start of first node has finished?


    
> Replication on startup not reliable
> -----------------------------------
>
>                 Key: ISPN-2495
>                 URL: https://issues.jboss.org/browse/ISPN-2495
>             Project: Infinispan
>          Issue Type: Bug
>          Components: State transfer
>    Affects Versions: 5.2.0.Beta3
>            Reporter: Thomas Fromm
>            Assignee: Mircea Markus
>            Priority: Blocker
>         Attachments: 2ND.log
>
>
> Situation: REPL cache with Cache Loader and ~500k entries and preload = true, and shared cache store.
> Problem: 1st node starts and the startup (until cache loaded all entries and so on) takes lets say 2min. When starting the 2nd node after e.g. 30s, then state is requested, but contains no results. cache loader is also not used, since its not the 1st node. Log from 2nd node is attached.
> So I've REPL cache over two (or more ) nodes, where only the 1st node contains the entries.
> When starting all other nodes later, then state transfer works and the data gets replicated.
> Maybe startup of all other nodes should be blocked until warm start of first node has finished?

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira