[jira] [Commented] (IGNITE-5849) Introduce ignite node persistent meta-store

["Konstantin Dudkov (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IGNITE-5849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16188260#comment-16188260 ] 

Konstantin Dudkov commented on IGNITE-5849:
-------------------------------------------

Ilya, thanks for your review

*     MetadataStorage - refactor, rename to IndexStorage and make subtype of MetaStorage.
renamed, but they now have completely different logic, so I find no way to make a relation
*     Validate that cache with name "MetaStorage" cannot be created with meaningful error message. Also validate memory policy name.
done
*     MetaStorage put/remove - can we avoid synchronization at this level? CacheDataStore works without it.
afaik CacheDataStore methods are called from CacheDataStore under the entry lock
*     MetaStorage.getOrAllocateMetas() - looks like copy/paste. Refactor?
*     MetaStorage.FreeListImpl.getRow(...) - looks like it does the same work as CacheDataRowAdapter.initFromLink(...). Reuse existing code, it already has fragmentation handling.
*     MetaStorage.start(...) - rename to init() or something else. Start(...) implicitly requires that stop(...) method also exists.
done
*     DataPageIO - reuse it for MetaStorage, skip unneeded fragments.
I put all common code to AbstractDataPageIO
*     GridCacheDatabaseSharedManager.getMetastoreData() - unused method? .start0() - commented code?
removed
*     IgniteCacheDatabaseSharedManager.addMemoryPolicy(...) and createPageEvictionTracker(...) - why protected?
they are called from outside



> Introduce ignite node persistent meta-store
> -------------------------------------------
>
>                 Key: IGNITE-5849
>                 URL: https://issues.apache.org/jira/browse/IGNITE-5849
>             Project: Ignite
>          Issue Type: New Feature
>          Components: persistence
>    Affects Versions: 2.1
>            Reporter: Alexey Goncharuk
>            Assignee: Konstantin Dudkov
>              Labels: important
>             Fix For: 2.4
>
>
> As persistence feature is being developed, we will have a need for a component, which reliably stores arbitrary metadata about node and cluster.
> We already have reserved partition IDs for this purpose, all we need to do is to extend the partition store abstraction to store non-cache objects and make sure this new store participates in all common recovery procedures.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)