[
https://issues.apache.org/jira/browse/HDFS-10178?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15201522#comment-15201522
]
Kihwal Lee commented on HDFS-10178:
-----------------------------------
For those who are wondering how critical this bug is: we hit this because of a
transient network issue, which caused multiple timeouts while writing the first
packet. After all original nodes were replaced, none of nodes in the pipeline
had the replica with a valid checksum type. The network issue was gone, but the
write could not continue because of the bug and permanently failed after 5
pipeline recovery attempts. I don't think it is common, but when it happens
the client cannot recover.
> Permanent write failures can happen if pipeline recoveries occur for the
> first packet
> -------------------------------------------------------------------------------------
>
> Key: HDFS-10178
> URL: https://issues.apache.org/jira/browse/HDFS-10178
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Kihwal Lee
> Assignee: Kihwal Lee
> Priority: Critical
> Attachments: HDFS-10178.patch, HDFS-10178.v2.patch
>
>
> We have observed that write fails permanently if the first packet doesn't go
> through properly and pipeline recovery happens. If the packet header is sent
> out, but the data portion of the packet does not reach one or more datanodes
> in time, the pipeline recovery will be done against the 0-byte partial block.
>
> If additional datanodes are added, the block is transferred to the new nodes.
> After the transfer, each node will have a meta file containing the header
> and 0-length data block file. The pipeline recovery seems to work correctly
> up to this point, but write fails when actual data packet is resent.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
