[
https://issues.apache.org/jira/browse/HDFS-927?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12828320#action_12828320
]
Todd Lipcon commented on HDFS-927:
----------------------------------
Ah, I see what you're saying. So, if you do a read that crosses a block
boundary A-B, and get 2 errors at the end of block A, and 2 errors at the start
of block B, you should still be OK?
I could go either way here. Part of me thinks that if you have errors on both
sides of a block boundary for a single read, your client is probably in a bad
state and you're likely to fail either way?
Since some are considering this an 0.20.2 blocker, could we get this commited
as a solid improvement over what's there now (which makes very little sense)
and then discuss whether the block boundary case should be improved?
> DFSInputStream retries too many times for new block locations
> -------------------------------------------------------------
>
> Key: HDFS-927
> URL: https://issues.apache.org/jira/browse/HDFS-927
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs client
> Affects Versions: 0.21.0, 0.22.0
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Critical
> Attachments: hdfs-927.txt
>
>
> I think this is a regression caused by HDFS-127 -- DFSInputStream is supposed
> to only go back to the NN max.block.acquires times, but in trunk it goes back
> twice as many - the default is 3, but I am counting 7 calls to
> getBlockLocations before an exception is thrown.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
