[
https://issues.apache.org/jira/browse/HDFS-1850?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Eli Collins updated HDFS-1850:
------------------------------
Attachment: hdfs-1850-7.patch
TestDatanodeBlockScanner passes for me in eclipse and when looped from the
command line. I think this is related to an earlier change. The error message
indicates we need to bump the number of attempts (ie it did see 1 corrupt
replica after 20 attempts, but it also bails on the 20th attempt).
{quote}
java.util.concurrent.TimeoutException: Timed out waiting for corrupt replicas.
Waiting for 1, but only found 1
{quote}
I'll bump the # attempts to 50 so we're more tolerant.
Updated patch attached.
> DN should transmit absolute failed volume count rather than increments to the
> NN
> --------------------------------------------------------------------------------
>
> Key: HDFS-1850
> URL: https://issues.apache.org/jira/browse/HDFS-1850
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: data-node, name-node
> Reporter: Eli Collins
> Assignee: Eli Collins
> Fix For: 0.22.0
>
> Attachments: hdfs-1850-1.patch, hdfs-1850-2.patch, hdfs-1850-3.patch,
> hdfs-1850-4.patch, hdfs-1850-5.patch, hdfs-1850-6.patch, hdfs-1850-7.patch
>
>
> The API added in HDFS-811 for the DN to report volume failures to the NN is
> "inc(DN)". However the given sequence of events will result in the NN
> forgetting about reported failed volumes:
> # DN loses a volume and reports it
> # NN restarts
> # DN re-registers to the new NN
> A more robust interface would be to have the DN report the total number of
> volume failures to the NN each heart beat (the same way other volume state is
> transmitted).
> This will likely be an incompatible change since it requires changing the
> Datanode protocol.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
