hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Enis Soztutar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-11571) Bulk load handling from secondary region replicas
Date Tue, 03 Mar 2015 00:14:05 GMT

    [ https://issues.apache.org/jira/browse/HBASE-11571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14344108#comment-14344108

Enis Soztutar commented on HBASE-11571:

nit: mutliple -> multiple. 

For the following, I think we should use HFileLinks to load the data. Both for exists check
and doing actual loading. The reason is that by the time the edit is replayed, the file can
be in archive directory already. That is why we should use HFileLinks always for finding the
+          List<String> storeFiles = storeDescriptor.getStoreFileList();
+          for (String storeFile : storeFiles) {
+            Path storeFilePath = new Path(familyPath, storeFile);
+            if (fs.getFileSystem().exists(storeFilePath)) {
+              store.bulkLoadHFile(storeFilePath);
You can use this: 
        StoreFileInfo storeFileInfo = fs.getStoreFileInfo(getColumnFamilyName(), compactionOutput);

Why is this needed? 
-        for (String storeFile : store.getStoreFileList()) {
-          assertTrue(storeFile.equals(storeFileNames.get(index++)));
-        }

You can use HTU.getDataTestDirOnTestFS(<random>) instead of this: 
+    TemporaryFolder testFolder = new TemporaryFolder();

This reader is not closed. You can use the local field (reader). 
+    WAL.Reader reader = createWALReaderForPrimary();

> Bulk load handling from secondary region replicas
> -------------------------------------------------
>                 Key: HBASE-11571
>                 URL: https://issues.apache.org/jira/browse/HBASE-11571
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Enis Soztutar
>            Assignee: Jeffrey Zhong
>         Attachments: HBASE-11571-rebase.patch, hbase-11571.patch
> We should be replaying the bulk load events from the primary region replica in the secondary
region replica so that the bulk loaded files will be made visible in the secondaries. 
> This will depend on HBASE-11567 and HBASE-11568

This message was sent by Atlassian JIRA

View raw message