[
https://issues.apache.org/jira/browse/HDFS-8319?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14573776#comment-14573776
]
Zhe Zhang commented on HDFS-8319:
---------------------------------
# Follow-on: {{divideOneStripe}} can be further simplified. The argument
{{rangeEndInBlockGroup}} is not necessary since it can be calculated from
{{rangeStartInBlockGroup}} and cell size.
# I think the current {{blockReaderOffsets}} logic works correctly. IIUC each
offset in the array represents the actual {{startOffset}} of the block reader
(looks like {{BlockReader}} should ideally have a getter for this internal
state?). For better readability, maybe we can just emulate how the block reader
offset evolves?
#* Is it possible that a failed or canceled request has also changed the
internal state of the block reader? If so we might {{skip}} too much.
#* Beside the above question, the main change I'd suggest is to add an assert
and increment the offset in {{StatefulStripeReader#updateState4SuccessRead}}:
{code}
Preconditions.checkState(blockReaderOffsets[result.index] ==
alignedStripe.getOffsetInBlock());
blockReaderOffsets[result.index] += alignedStripe.getSpanInBlock();
{code}
#* Nit: in {{prepareBlockReader}}, {{blockReaderOffsets\[i\] =
alignedStripe.getOffsetInBlock()}} can be changed to {{offsetInBlock}} to be
consistent with the {{getBlockReaderWithRetry}} call.
# {{prepareBlockReader}} can be {{prepareParityBlockReader}}?
> Erasure Coding: support decoding for stateful read
> --------------------------------------------------
>
> Key: HDFS-8319
> URL: https://issues.apache.org/jira/browse/HDFS-8319
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Jing Zhao
> Assignee: Jing Zhao
> Attachments: HDFS-8319.001.patch, HDFS-8319.002.patch,
> HDFS-8319.003.patch, HDFS-8319.004.patch
>
>
> HDFS-7678 adds the decoding functionality for pread. This jira plans to add
> decoding to stateful read.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
