[
https://issues.apache.org/jira/browse/HDFS-13522?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17691455#comment-17691455
]
Takanobu Asanuma commented on HDFS-13522:
-----------------------------------------
[~simbadzina] Thanks for your comment.
For clientC(2.7.x), in my test case, it isn't small changes because it needs
the latest changes and HDFS-12943. The changes may be small for clientB(3.3.4),
but we still have to apply the latest patches to all these clients who want to
send read requests to observers behind routers. Upgrading all clients is
painful in most environments.
In this jira, there are Design A and Design B. Design A is effective because it
saves msycn calls but needs to upgrade the client side since it extends the RPC
header. Design B is that the router always calls msync for each read, which is
expensive, but old clients can send read requests to the observer. So, the
combination of Design A (for new clients) and Design B (for old clients) was
chosen. Am I right?
I'm not sure why Design B is finally not implemented in the current
implementation (is there any discussion about it?). I think most users still
need Design B, as others mentioned in this jira. Is it possible to implement an
option(configuration) on the Router side to switch the behavior between Design
A and Design B for old clients?
> HDFS-13522: Add federated nameservices states to client protocol and
> propagate it between routers and clients.
> --------------------------------------------------------------------------------------------------------------
>
> Key: HDFS-13522
> URL: https://issues.apache.org/jira/browse/HDFS-13522
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: federation, namenode
> Reporter: Erik Krogen
> Assignee: Simbarashe Dzinamarira
> Priority: Major
> Labels: pull-request-available
> Fix For: 3.4.0
>
> Attachments: HDFS-13522.001.patch, HDFS-13522.002.patch,
> HDFS-13522_WIP.patch, RBF_ Observer support.pdf, Router+Observer RPC
> clogging.png, ShortTerm-Routers+Observer.png,
> observer_reads_in_rbf_proposal_simbadzina_v1.pdf,
> observer_reads_in_rbf_proposal_simbadzina_v2.pdf
>
> Time Spent: 20h 50m
> Remaining Estimate: 0h
>
> Changes will need to occur to the router to support the new observer node.
> One such change will be to make the router understand the observer state,
> e.g. {{{}FederationNamenodeServiceState{}}}.
> This patch captures the state of all namespaces in the routers and propagates
> it to clients. A follow up patch will change router behavior to direct
> requests to the observer.
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
