[
https://issues.apache.org/jira/browse/SOLR-10007?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Erick Erickson updated SOLR-10007:
----------------------------------
Attachment: SOLR-10007.patch
Final patch with CHANGES entry.
I've tested this pretty thoroughly, all tests pass 20 iterations of the full
unit tests.
This is a big patch, but it has a lot of mechanical changes, i.e.
core.getCoreDescriptor().getCoreContainer()
is now just
core.getCoreContainer()
Significant fragility in the code was because we kept CoreDescriptors in
multiple places, which is no longer the case.
I also renamed a couple of the internal methods to make their purpose clearer.
I also added a "reload" method to CoresLocator, there has to be the possibility
of re-reading the coreDescriptor from disk.
> Clean up references to CoreContainer and CoreDescriptors
> --------------------------------------------------------
>
> Key: SOLR-10007
> URL: https://issues.apache.org/jira/browse/SOLR-10007
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Affects Versions: trunk, 6.4
> Reporter: Erick Erickson
> Assignee: Erick Erickson
> Attachments: SOLR-10007.patch, SOLR-10007.patch, SOLR-10007.patch,
> SOLR-10007.patch
>
>
> It's a bit weird that CoreDescriptor contains a reference to CoreContainer.
> It seems like the SolrCore should reference CoreContainer.
> Similarly, SolrCore keeps a copy of CoreDescriptor and another copy is kept
> in the various CoreDescirptor lists making it difficult to track which is the
> "real" CoreDescriptor.
> This is an umbrella issue as I think this (and perhaps other) issues can be
> tackled separately.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]