[
https://issues.apache.org/jira/browse/OAK-3449?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Julian Reschke updated OAK-3449:
--------------------------------
Attachment: cid.diff
As an exercise, I changed DocumentNodeStore to add 1000 to the cluster id (when
!= 0).
Result: had to do 2 trivial fixes in test cases. One test
(oak-core/src/test/java/org/apache/jackrabbit/oak/plugins/document/LastRevRecoveryRandomizedIT)
continues to fail because it relies on the ability to predict the clusterId,
we probably would have to rewrite it.
In summary: this (making the clusterId != 0 just a hint) looks promising.
> DocumentNodeStore support for predefined clusterIds should use
> ClusterNodeInfos
> -------------------------------------------------------------------------------
>
> Key: OAK-3449
> URL: https://issues.apache.org/jira/browse/OAK-3449
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: core, mongomk, rdbmk
> Reporter: Julian Reschke
> Assignee: Julian Reschke
> Labels: test
> Attachments: cid.diff
>
>
> We currently support explicit assignment of clusterId
> (builder.setClusterId()). In this case, DocumentNodeStore uses the specified
> cluster id and skips all code related to creating/maintaining
> ClusterNodeInfos.
> This feature is mainly (only?) used for testing (mainly allowing multiple
> instances to run from the same machine/instance combination). This works, but
> causes the logic related to ClusterNodeInfo not to be used at all (for
> instance, LastRevRecovery).
> So we ought to change this config option to use ClusterNodeInfo in a way that
> is at least similar to real-world use.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
