[
https://issues.apache.org/jira/browse/HDFS-13767?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16567060#comment-16567060
]
Erik Krogen commented on HDFS-13767:
------------------------------------
Cool, thanks for incorporating the feedback Chen! I agree with Konstantin's
comments and have a few more to add:
* The comment at the top of {{GlobalStateIdContext#receiveRequestState()}} is a
little confusing. I agree that we don't need to care about the client state in
this case, but shouldn't we still return {{header.getStateId()}} to meet the
interface contract? Is there any disadvantage to doing so?
* I'm not a fan of the changes within {{Server#run()}}; they are extremely
HDFS-specific in a Hadoop-general class. It seems it would be better if there
was a way for the {{AlignmentContext}}, a generic/pluggable component, to make
the decision about how to handle the call. Something like a new method
{{boolean shouldDeferProcessing(Server.Call call)}}
* GlobalStateIdContext L81/82, you shouldn't use a string comparison to compare
an enum, you can use {{FSNamesystem#getState()}} to retrieve the enum directly.
* In {{TestObserverNode}}, you shouldn't use a star-import for {{Assert}}.
Also, this is nitpicky, but the {{AtomicBoolean}} import should probably be
grouped with the other {{java.}} imports.
* Why did you move {{setUpCluster()}} out of {{TestObserverNode#setUp()}} and
into the individual test methods? It seems like code duplication without any
advantage. Let me know if I'm missing something.
* You can make the tests much faster by enabling in-progress edit log tailing
which will use the new fast-path from HDFS-13150 :) Just use configs like:
{code}
conf.setBoolean(DFSConfigKeys.DFS_HA_TAILEDITS_INPROGRESS_KEY, true);
conf.setTimeDuration(DFS_HA_TAILEDITS_PERIOD_KEY, 100,
TimeUnit.MILLISECONDS);
{code}
Note that you should use the {{setTimeDuration}} for time-related configs these
days; else you get messages like:
{code}INFO Configuration.deprecation (Configuration.java:logDeprecation(1395))
- No unit for dfs.ha.log-roll.period(60) assuming SECONDS{code}
> Add msync server implementation.
> --------------------------------
>
> Key: HDFS-13767
> URL: https://issues.apache.org/jira/browse/HDFS-13767
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: namenode
> Reporter: Chen Liang
> Assignee: Chen Liang
> Priority: Major
> Attachments: HDFS-13767-HDFS-12943.001.patch,
> HDFS-13767.WIP.001.patch, HDFS-13767.WIP.002.patch, HDFS-13767.WIP.003.patch,
> HDFS-13767.WIP.004.patch
>
>
> This is a followup on HDFS-13688, where msync API is introduced to
> {{ClientProtocol}} but the server side implementation is missing. This is
> Jira is to implement the server side logic.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
