[
https://issues.apache.org/jira/browse/HDDS-675?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16676455#comment-16676455
]
Mukul Kumar Singh commented on HDDS-675:
----------------------------------------
Thanks for working on this [~shashikant]. The patch generally looks good to me.
Some comments.
1) Can the ioException in ChunkOutputStream be saved and thrown later when
watchForCommit is issued on block close.
2) TestCloseContainerHandlingByClient:121-124, these lines are commented, there
are lots of commented lines of code,
can you please take remove those
3) Failures of TestMultipleContainerReadWrite seems related to the patch.
Future Patches:
a) ChunkGroupOutputStream:191, the buffer size here is the same as blockSize,
will this be changed to maxBufferSize in one of the followup patches ? This can
be done by maintaining buffer as list of smallers buffers.
b) Currently with this patch, the retry is for TimeoutException as well as
ClosedContainerException, I feel we should add policies on which exceptions
should be retried.
c) Also the code to oppurtunistically discard the buffer, most of the code
inside this can be encapsulated inside Ratis. Ratis client can maintain the
commitIndexes for each client and then this information can be used by
ChunkOutPutStream to opportunistically discard buffers.
> Add blocking buffer and use watchApi for flush/close in OzoneClient
> -------------------------------------------------------------------
>
> Key: HDDS-675
> URL: https://issues.apache.org/jira/browse/HDDS-675
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: Ozone Client
> Reporter: Shashikant Banerjee
> Assignee: Shashikant Banerjee
> Priority: Major
> Attachments: HDDS-675.000.patch, HDDS-675.001.patch
>
>
> For handling 2 node failures, a blocking buffer will be used which will wait
> for the flush commit index to get updated on all replicas of a container via
> Ratis.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]