[
https://issues.apache.org/jira/browse/HADOOP-1134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12486740
]
Raghu Angadi commented on HADOOP-1134:
--------------------------------------
The existing protocol and the one proposed in the previous comment don't have
confirmation from the receiver that complete block was received. Looks like it
is necessary for sender to be confident that the block was properly received.
Extention to the above :
In response to OP_WRITE_BLOCK, receiver sends one of the following one byte
code to the sender :
----------------------------
| 1 byte Write status |
----------------------------
Status is one of the following :
OP_STATUS_SUCCESS
OP_STATUS_ERROR_CHECKSUM ( currently not used )
OP_STATUS_ERROR ( Misc error. e.g. offset mismatch for successive DATA_CHUNKS )
OP_STATUS_EXISTS
OP_STATUS_EXISTS could be used when client try to resend the block since it
failed to receive the 'OP_STATUS_OK' response from previous attempt. Sometimes
this could be an error. The receiver could close the connection anytime and it
could write status byte before closing the connection.
> Block level CRCs in HDFS
> ------------------------
>
> Key: HADOOP-1134
> URL: https://issues.apache.org/jira/browse/HADOOP-1134
> Project: Hadoop
> Issue Type: New Feature
> Components: dfs
> Reporter: Raghu Angadi
> Assigned To: Raghu Angadi
>
> Currently CRCs are handled at FileSystem level and are transparent to core
> HDFS. See recent improvement HADOOP-928 ( that can add checksums to a given
> filesystem ) regd more about it. Though this served us well there a few
> disadvantages :
> 1) This doubles namespace in HDFS ( or other filesystem implementations ). In
> many cases, it nearly doubles the number of blocks. Taking namenode out of
> CRCs would nearly double namespace performance both in terms of CPU and
> memory.
> 2) Since CRCs are transparent to HDFS, it can not actively detect corrupted
> blocks. With block level CRCs, Datanode can periodically verify the checksums
> and report corruptions to namnode such that name replicas can be created.
> We propose to have CRCs maintained for all HDFS data in much the same way as
> in GFS. I will update the jira with detailed requirements and design. This
> will include same guarantees provided by current implementation and will
> include a upgrade of current data.
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
