1
0

cleaner should now use commit timeline and not include deltacomits

This commit is contained in:
Nishith Agarwal
2018-12-26 13:37:22 -08:00
committed by vinoth chandar
parent 68723764ed
commit 994d42d307
9 changed files with 31 additions and 24 deletions

View File

@@ -493,7 +493,7 @@ public class TestAsyncCompaction extends TestHoodieClientBase {
private List<HoodieDataFile> getCurrentLatestDataFiles(HoodieTable table, HoodieWriteConfig cfg) throws IOException {
FileStatus[] allFiles = HoodieTestUtils.listAllDataFilesInPath(table.getMetaClient().getFs(), cfg.getBasePath());
HoodieTableFileSystemView
view = new HoodieTableFileSystemView(table.getMetaClient(), table.getCompletedCommitTimeline(), allFiles);
view = new HoodieTableFileSystemView(table.getMetaClient(), table.getCompletedCommitsTimeline(), allFiles);
List<HoodieDataFile> dataFilesToRead = view.getLatestDataFiles().collect(Collectors.toList());
return dataFilesToRead;
}

View File

@@ -126,8 +126,8 @@ public class TestCleaner extends TestHoodieClientBase {
// Should have 100 records in table (check using Index), all in locations marked at commit
HoodieTable table = HoodieTable.getHoodieTable(metaClient, getConfig(), jsc);
assertFalse(table.getCompletedCommitTimeline().empty());
String commitTime = table.getCompletedCommitTimeline().getInstants().findFirst().get().getTimestamp();
assertFalse(table.getCompletedCommitsTimeline().empty());
String commitTime = table.getCompletedCommitsTimeline().getInstants().findFirst().get().getTimestamp();
assertFalse(table.getCompletedCleanTimeline().empty());
assertEquals("The clean instant should be the same as the commit instant", commitTime,
table.getCompletedCleanTimeline().getInstants().findFirst().get().getTimestamp());
@@ -380,7 +380,7 @@ public class TestCleaner extends TestHoodieClientBase {
HoodieTableMetaClient metadata = new HoodieTableMetaClient(jsc.hadoopConfiguration(), basePath);
HoodieTable table1 = HoodieTable.getHoodieTable(metadata, cfg, jsc);
HoodieTimeline activeTimeline = table1.getCompletedCommitTimeline();
HoodieTimeline activeTimeline = table1.getCompletedCommitsTimeline();
Optional<HoodieInstant> earliestRetainedCommit = activeTimeline.nthFromLastInstant(maxCommits - 1);
Set<HoodieInstant> acceptableCommits = activeTimeline.getInstants().collect(Collectors.toSet());
if (earliestRetainedCommit.isPresent()) {

View File

@@ -176,7 +176,7 @@ public class TestMergeOnReadTable {
Stream<HoodieDataFile> dataFilesToRead = roView.getLatestDataFiles();
assertTrue(!dataFilesToRead.findAny().isPresent());
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFilesToRead = roView.getLatestDataFiles();
assertTrue("RealtimeTableView should list the parquet files we wrote in the delta commit",
dataFilesToRead.findAny().isPresent());
@@ -210,7 +210,7 @@ public class TestMergeOnReadTable {
client.compact(compactionCommitTime);
allFiles = HoodieTestUtils.listAllDataFilesInPath(dfs, cfg.getBasePath());
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFilesToRead = roView.getLatestDataFiles();
assertTrue(dataFilesToRead.findAny().isPresent());
@@ -283,7 +283,7 @@ public class TestMergeOnReadTable {
Stream<HoodieDataFile> dataFilesToRead = roView.getLatestDataFiles();
assertTrue(!dataFilesToRead.findAny().isPresent());
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFilesToRead = roView.getLatestDataFiles();
assertTrue("RealtimeTableView should list the parquet files we wrote in the delta commit",
dataFilesToRead.findAny().isPresent());
@@ -320,7 +320,7 @@ public class TestMergeOnReadTable {
assertFalse(commit.isPresent());
allFiles = HoodieTestUtils.listAllDataFilesInPath(dfs, cfg.getBasePath());
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFilesToRead = roView.getLatestDataFiles();
assertTrue(dataFilesToRead.findAny().isPresent());
@@ -380,7 +380,7 @@ public class TestMergeOnReadTable {
HoodieTable hoodieTable = HoodieTable.getHoodieTable(metaClient, cfg, jsc);
FileStatus[] allFiles = HoodieTestUtils.listAllDataFilesInPath(metaClient.getFs(), cfg.getBasePath());
HoodieTableFileSystemView roView = new HoodieTableFileSystemView(metaClient,
hoodieTable.getCompletedCommitTimeline(), allFiles);
hoodieTable.getCompletedCommitsTimeline(), allFiles);
final String absentCommit = newCommitTime;
assertFalse(roView.getLatestDataFiles().filter(file -> {
@@ -430,7 +430,7 @@ public class TestMergeOnReadTable {
Stream<HoodieDataFile> dataFilesToRead = roView.getLatestDataFiles();
assertTrue(!dataFilesToRead.findAny().isPresent());
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFilesToRead = roView.getLatestDataFiles();
assertTrue("RealtimeTableView should list the parquet files we wrote in the delta commit",
dataFilesToRead.findAny().isPresent());
@@ -504,7 +504,7 @@ public class TestMergeOnReadTable {
metaClient = new HoodieTableMetaClient(jsc.hadoopConfiguration(), cfg.getBasePath());
hoodieTable = HoodieTable.getHoodieTable(metaClient, cfg, jsc);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFiles = roView.getLatestDataFiles().map(hf -> hf.getPath()).collect(Collectors.toList());
recordsRead = HoodieMergeOnReadTestUtils.getRecordsUsingInputFormat(dataFiles, basePath);
// check that the number of records read is still correct after rollback operation
@@ -599,7 +599,7 @@ public class TestMergeOnReadTable {
Map<String, Long> parquetFileIdToSize = dataFilesToRead.collect(
Collectors.toMap(HoodieDataFile::getFileId, HoodieDataFile::getFileSize));
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitTimeline(), allFiles);
roView = new HoodieTableFileSystemView(metaClient, hoodieTable.getCompletedCommitsTimeline(), allFiles);
dataFilesToRead = roView.getLatestDataFiles();
List<HoodieDataFile> dataFilesList = dataFilesToRead.collect(Collectors.toList());
assertTrue("RealtimeTableView should list the parquet files we wrote in the delta commit",