[
https://issues.apache.org/jira/browse/HDFS-13300?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16408201#comment-16408201
]
Elek, Marton commented on HDFS-13300:
-------------------------------------
Thanks to work on this [~nandakumar131]. I am very happy to see that the
hdsl/ozone became more independent from datanode/hdfs.
Overall the patch looks good, I have some comments:
1. I am not sure if we need infoPort. As I remember it is used for DatanodeHttp
server which is not required for hdsl/ozone any more. (We need an addition
server in hdsl to help the container replication which could be added to the
descriptor but it can be part of HDFS-11686)
2. You renamed HdslServerPlugin to HdslDatanodeService. I am happy with this
change but please keep the two plugin names consistent: in this case please
rename ObjectStoreRestPlugin to ObjectStoreDatanodeService (or
OzoneDatanodeService). As the two classes are very similar I prefer to use
similar names (Especially as the names are used in the configuration).
3. I can't see how the race condition between hdsl/object store
services(=plugins) are handled. As I see the ObjectStoreRestPlugin updates the
ozoneRestPort in the one DatanodeDetails which is part of the scm--datanode.
heartbeat. But I think the SCMNodeManager.hadleHeartbeat should contain some
logic to update the ports from the heartbeat (I think it's used from the
discovery). I think this is the reason behind the failing REST related unit
tests (didn't check, just my guess).
4. I am not sure if we need to persist the DatanodeDetails. I think it's enogh
to persiste the UUID. The ports could be changed (and in containerized
environment host/ip also could be changed in case of moving a container to a
different host). But I understand this was not changed by this patch.
I think the whole logic should work even if I modify any of the ports. If I
understood well, now it is true, as all the ports are updated after reading the
datanode descriptor from the file.
5. This is a smal one, but some javadoc still use the "Datanode ID" expression
which could be confusing (StorageContainerDatanodeProtocol, SCMTestMock)
> Ozone: Remove DatanodeID dependency from HDSL and Ozone
> --------------------------------------------------------
>
> Key: HDFS-13300
> URL: https://issues.apache.org/jira/browse/HDFS-13300
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: ozone
> Reporter: Nanda kumar
> Assignee: Nanda kumar
> Priority: Major
> Attachments: HDFS-13300-HDFS-7240.000.patch,
> HDFS-13300-HDFS-7240.001.patch
>
>
> DatanodeID has been modified to add HDSL/Ozone related information
> previously. This jira is to remove DatanodeID dependency from HDSL/Ozone to
> make it truly pluggable without having the need to modify DatanodeID.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
