hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "kanaka kumar avvaru (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-8661) DataNode should filter the set of NameSpaceInfos passed to Datasets
Date Tue, 30 Jun 2015 12:42:05 GMT

    [ https://issues.apache.org/jira/browse/HDFS-8661?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14608223#comment-14608223

kanaka kumar avvaru commented on HDFS-8661:

Patch looks good [~arpitagarwal], Can you please check if the below comments are valid my
from view

1) DataNode#refreshVolumes
          // Find the storage services matching the NodeType for this dataset.
          final List<NamespaceInfo> filteredNsInfos = new ArrayList<>();
          for (final NamespaceInfo nsInfo : nsInfos) {
            if (datasets.get(nsInfo.getNodeType()) == dataset) {
            } else {
              LOG.error("arpit: Filtered out nsInfo " + nsInfo);

As per my understanding at present every NS should have a dataset mapped. So, should this
logic mark an IOException for failed case instead of just log error?
Small nit: You may want to remove text {{arpit:}} and dataset identify in the log

{code} LOG.info("FsDatasetImpl: Adding volume " + location.getFile());{code}

shall we improve the logging with {{nsInfos}} also to trace that all the block pools from
Ozone & File system registered on this volume

> DataNode should filter the set of NameSpaceInfos passed to Datasets
> -------------------------------------------------------------------
>                 Key: HDFS-8661
>                 URL: https://issues.apache.org/jira/browse/HDFS-8661
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: datanode
>    Affects Versions: HDFS-7240
>            Reporter: Arpit Agarwal
>            Assignee: Arpit Agarwal
>         Attachments: HDFS-8661-HDFS-7240.01.patch, HDFS-8661-HDFS-7240.02.patch, HDFS-8661-HDFS-7240.03.patch
> {{DataNode#refreshVolumes}} passes the list of NamespaceInfos to each dataset when adding
new volumes.
> This list should be filtered by the correct NodeType(s) for each dataset. e.g. in a shared
HDFS+Ozone cluster, FsDatasets would be notified of NN block pools and Ozone datasets would
be notified of Ozone block pool(s).

This message was sent by Atlassian JIRA

View raw message