[
https://issues.apache.org/jira/browse/HDFS-15255?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17084043#comment-17084043
]
Stephen O'Donnell commented on HDFS-15255:
------------------------------------------
We still need to address the Find Bugs warning somehow. I think we should
revert the earlier change I suggested, as that did not help:
{code}
for (DatanodeInfo loc : block.getLocations()) {
to
for (DatanodeInfoWithStorage loc : block.getLocations()) {
{code}
Then try casting the datanode variable to a DatanodeInfo in this line:
{code}
if (loc.equals(datanode)) {
{code}
For the log message added, we should ideally mention the actual parameter names
rather than the local variable names as that would give the user a better idea
of what is wrong.
This log message will be logged for every getBlockLocations call I think. While
it highlights a problem, and would only be logged if there is a problem, it
would create a lot of noise in the logs. In saying that, if we only log it
once, then it may be missed. I think I would prefer to log it just once to
avoid the log noise. We could do that in the DatanodeManager constructor with a
message like:
{code}
LOG.warn("{} and {} are incompatible and only one can be enabled. Both are
currently enabled.",
DFSConfigKeys.DFS_NAMENODE_READ_CONSIDERSTORAGETYPE_KEY,
DFSConfigKeys.DFS_NAMENODE_READ_CONSIDERLOAD_KEY
{code}
There is also one checkstyle warning for an unused import.
> Consider StorageType when DatanodeManager#sortLocatedBlock()
> ------------------------------------------------------------
>
> Key: HDFS-15255
> URL: https://issues.apache.org/jira/browse/HDFS-15255
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Lisheng Sun
> Assignee: Lisheng Sun
> Priority: Major
> Fix For: 3.4.0
>
> Attachments: HDFS-15255.001.patch, HDFS-15255.002.patch,
> HDFS-15255.003.patch, HDFS-15255.004.patch
>
>
> When only one replica of a block is SDD, the others are HDD.
> When the client reads the data, the current logic is that it considers the
> distance between the client and the dn. I think it should also consider the
> StorageType of the replica. Priority to return fast StorageType node when the
> distance is same.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
