[ https://issues.apache.org/jira/browse/HDFS-13421?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16537720#comment-16537720 ]
Virajith Jalaparti edited comment on HDFS-13421 at 7/9/18 10:35 PM: -------------------------------------------------------------------- bq. So we chose a design where the NN is quite explicit in what needs to happen the DN is a dumb worker. Sure, I am fine with concatenating multiple blocks on the DN. However, my concern with the current implementation is that it reads all blocks into in-memory buffers and writes them to the remote. This should be done without having to read all the data into memory. bq. If you think it's best, I can try to add some of the NN side things. I meant to connect {{SyncServiceSatisfierDatanodeWorker#processSyncTasks}} with {{BPOfferService#processCommandFromActive}} and complete the DN side of the implementation in this JIRA. The NN functionality can be handled as a separate sub-task. was (Author: virajith): bq. So we chose a design where the NN is quite explicit in what needs to happen the DN is a dumb worker. Sure, I am fine with concatenating multiple blocks on the DN. However, my concern with the current implementation is that it reads all blocks into in-memory buffers and writes them to the remote. This should be done without having to read all the data into memory. bq. If you think it's best, I can try to add some of the NN side things. I meant to connect {{SyncServiceSatisfierDatanodeWorker#processSyncTasks}} with {{BPOfferService#processCommandFromActor}} and complete the DN side of the implementation in this JIRA. The NN functionality can be handled as a separate sub-task. > [PROVIDED Phase 2] Implement DNA_BACKUP command in Datanode > ----------------------------------------------------------- > > Key: HDFS-13421 > URL: https://issues.apache.org/jira/browse/HDFS-13421 > Project: Hadoop HDFS > Issue Type: Sub-task > Reporter: Ewan Higgs > Assignee: Ewan Higgs > Priority: Major > Attachments: HDFS-13421-HDFS-12090.001.patch, > HDFS-13421-HDFS-12090.002.patch, HDFS-13421-HDFS-12090.003.patch > > > HDFS-13310 introduces an API for DNA_BACKUP. Here, we implement DNA_BACKUP > command in Datanode. > These have been broken up to make reviewing it easier. -- 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