[
https://issues.apache.org/jira/browse/HADOOP-1134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12511620
]
Hairong Kuang commented on HADOOP-1134:
---------------------------------------
When I test Block-level crc upgrade, I set a file's replication factor to be 4.
Afterwards I manually corrupted 3 out of 4 replicas of the crc file and then
upgraded it to block-level crc dfs. Although there is one correct crc replica
before upgrade, the current block upgrade algothrim does not handle the case
well and reading the file gets a ChecksumError after the upgrade.
It would be nice if block upgrade keeps all the crc replicas instead of taking
the majority replicas when there is a crc mismatch. For example, for data block
A, if we have replicas A1, A2, and A3, and crc replicas CRC1, CRC2, CRC3, then
we can have Ai creates its meta file by reading CRCi (i=1,2,3). To form a
one-to-one mapping from replica Ai to CRCi, we can sort all data replicas and
sort all its crc replicas respectively.
> 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
> Assignee: Raghu Angadi
> Attachments: BlockLevelCrc-07032007.patch,
> BlockLevelCrc-07052007.patch, BlockLevelCrc-07062007.patch,
> BlockLevelCrc-07062007.patch, BlockLevelCrc-07062007.patch,
> DfsBlockCrcDesign-05305007.htm, readBuffer.java
>
>
> 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.
