hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Walter Su (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-9180) Update excluded DataNodes in DFSStripedOutputStream based on failures in data streamers
Date Sun, 04 Oct 2015 13:14:26 GMT

    [ https://issues.apache.org/jira/browse/HDFS-9180?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14942662#comment-14942662
] 

Walter Su commented on HDFS-9180:
---------------------------------

1. After calling replaceFailedStreamers(), does getExcludedNodes() always return empty? //
allocateNewBlock()

bq. When replacing a StripedDataStreamer, should also reset the currentPacket to null if the
streamer is the first one.
2. You assumed replaceFailedStreamers() will call when currentIdx==0. That's true. It's just
I think if we change
{code}
-  if (i == 0) {
+  if (i == getCurrentIndex()) {
{code}
it's more readable. And getCurrentIndex() can get index from {{currentBlockGroup}} instead
of {{streamer}}.

3. I don't know why #7 streamer didn't bump gs in the failed test. I run 40min and can't reproduce.
If we check blockGroup.numBytes() instead of DATA_STREAMING, and assert stage is DATA_STREAMING,
does it help debug if test fails again?
{code}
if (streamer.isHealthy() &&
    // the streamer may not be in STREAMING stage if the block length is
    // less than a stripe
    streamer.getStage() == BlockConstructionStage.DATA_STREAMING) {
  streamer.setExternalError();
  healthySet.add(streamer);
}
{code}
#2, #3 not related.

> Update excluded DataNodes in DFSStripedOutputStream based on failures in data streamers
> ---------------------------------------------------------------------------------------
>
>                 Key: HDFS-9180
>                 URL: https://issues.apache.org/jira/browse/HDFS-9180
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: erasure-coding
>    Affects Versions: 3.0.0
>            Reporter: Jing Zhao
>            Assignee: Jing Zhao
>         Attachments: HDFS-9180.000.patch, HDFS-9180.001.patch, HDFS-9180.002.patch
>
>
> This is a TODO in HDFS-9040: based on the failures all the striped data streamers hit,
the DFSStripedOutputStream should keep a record of all the DataNodes that should be excluded.
> This jira will also fix several bugs in the DFSStripedOutputStream. Will provide more
details in the comment.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message