[
https://issues.apache.org/jira/browse/HDFS-10881?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15946228#comment-15946228
]
Chris Douglas commented on HDFS-10881:
--------------------------------------
* If the create/mod time needs to be set, can it at least use the same
timestamp? That call can be expensive
* If {{remove(T)}} is called on a record that isn't in the state store, is the
result successful? Please update the javadoc
* If an implementation should maintain invariant properties (atomicity,
idempotence, etc.) for these calls, it might be worth calling those out. For
example, if {{putAll}} fails partway through, should the state store guarantee
rollback, or does the client just skip collisions? I don't know if the
annotations added in HDFS-4974 are still used, or apply to the state store
implementations this supports.
** The API is very generic. In context with the implementation, it may be
possible to simplify it.
* Walking the type hierarchy in {{StateStoreUtils#getRecordClass}} by classname
is unfortunate. Since it'd be cross-cutting across subsequent patches we can
revisit it later.
+1 for committing to the branch
> Federation State Store Driver API
> ---------------------------------
>
> Key: HDFS-10881
> URL: https://issues.apache.org/jira/browse/HDFS-10881
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: fs
> Reporter: Jason Kace
> Assignee: Jason Kace
> Attachments: HDFS-10881-HDFS-10467-001.patch,
> HDFS-10881-HDFS-10467-002.patch, HDFS-10881-HDFS-10467-003.patch,
> HDFS-10881-HDFS-10467-004.patch, HDFS-10881-HDFS-10467-005.patch,
> HDFS-10881-HDFS-10467-006.patch, HDFS-10881-HDFS-10467-007.patch,
> HDFS-10881-HDFS-10467-008.patch, HDFS-10881-HDFS-10467-009.patch,
> HDFS-10881-HDFS-10467-010.patch, HDFS-10881-HDFS-10467-011.patch,
> HDFS-10881-HDFS-10467-012.patch
>
>
> The API interfaces and minimal classes required to support a state store data
> backend such as ZooKeeper or a file system.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
