[ https://issues.apache.org/jira/browse/HDFS-12976?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16508426#comment-16508426 ]
Chao Sun commented on HDFS-12976: --------------------------------- Thanks [~xkrogen], [~vagarychen] and [~shv], for the very helpful comments. [~xkrogen]: I think the {{ThreadLocal}} is just a minor optimization and I don't have strong preference over either. Regarding the second comment, yes a slow node could increase the startup/failover time when it initializes the proxy list, but won't this happen anyway with the old approach where the slowness would move to the normal RPC call and affect the client? [~vagarychen] besides what Erik pointed out, we can potentially add more logic to {{invoke}} methods, where it can check the {{LastSeenId}} of both ANN and SBN/Observer, before forwarding the request (although I'm not sure if the {{LastSeenId}} is accessible from the proxy class). It doesn't seem like the current approach is in conflict with what you intend to do. Will address the comments from Konstantin and let's move discussions to the follow up JIRAs. > Introduce ObserverReadProxyProvider > ----------------------------------- > > Key: HDFS-12976 > URL: https://issues.apache.org/jira/browse/HDFS-12976 > Project: Hadoop HDFS > Issue Type: Sub-task > Components: hdfs-client > Reporter: Konstantin Shvachko > Assignee: Chao Sun > Priority: Major > Attachments: HDFS-12976-HDFS-12943.000.patch, HDFS-12976.WIP.patch > > > {{StandbyReadProxyProvider}} should implement {{FailoverProxyProvider}} > interface and be able to submit read requests to ANN and SBN(s). -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org