@Test
public void testEviction4() throws Exception {
try {
Configuration conf = cluster.getConf();
FileSystem fs = FileSystem.get(conf);
fs.delete(new Path("/"), true);
stream.clear();
Pair<List<String>, List<String>> pair = createTestData("/data");
FeedEvictor.main(
new String[] {
"-feedBasePath",
LocationType.DATA.name()
+ "="
+ cluster.getConf().get(HadoopClientFactory.FS_DEFAULT_NAME_KEY)
+ "/data/YYYY/feed3/dd/MM/?{MONTH}/more/?{HOUR}",
"-retentionType", "instance",
"-retentionLimit", "months(5)",
"-timeZone", "UTC",
"-frequency", "hourly",
"-logFile",
conf.get(HadoopClientFactory.FS_DEFAULT_NAME_KEY)
+ "/falcon/staging/feed/2012-01-01-04-00",
"-falconFeedStorageType", Storage.TYPE.FILESYSTEM.name(),
});
Assert.assertEquals("instances=NULL", stream.getBuffer());
stream.clear();
String dataPath = "/data/YYYY/feed4/dd/MM/02/more/hello";
String logFile = hdfsUrl + "/falcon/staging/feed/instancePaths-2012-01-01-02-00.csv";
FeedEvictor.main(
new String[] {
"-feedBasePath",
LocationType.DATA.name()
+ "="
+ cluster.getConf().get(HadoopClientFactory.FS_DEFAULT_NAME_KEY)
+ dataPath,
"-retentionType", "instance",
"-retentionLimit", "hours(5)",
"-timeZone", "UTC",
"-frequency", "hourly",
"-logFile", logFile,
"-falconFeedStorageType", Storage.TYPE.FILESYSTEM.name(),
});
Assert.assertEquals("instances=NULL", stream.getBuffer());
Assert.assertEquals(readLogFile(new Path(logFile)), getExpectedInstancePaths(dataPath));
assertFailures(fs, pair);
} catch (Exception e) {
Assert.fail("Unknown exception", e);
}
}
@Test(enabled = false)
public void testEvictionWithEmptyDirs() throws Exception {
try {
Configuration conf = cluster.getConf();
FileSystem fs = FileSystem.get(conf);
fs.delete(new Path("/"), true);
stream.clear();
Pair<List<String>, List<String>> pair =
generateInstances(
fs, "feed1", "yyyy/MM/dd/'more'/yyyy", 10, TimeUnit.DAYS, "/data", false);
final String storageUrl = cluster.getConf().get(HadoopClientFactory.FS_DEFAULT_NAME_KEY);
String dataPath =
LocationType.DATA.name()
+ "="
+ storageUrl
+ "/data/YYYY/feed1/mmHH/dd/MM/?{YEAR}/?{MONTH}/?{DAY}/more/?{YEAR}";
String logFile = hdfsUrl + "/falcon/staging/feed/instancePaths-2012-01-01-01-00.csv";
long beforeDelCount =
fs.getContentSummary(new Path(("/data/YYYY/feed1/mmHH/dd/MM/"))).getDirectoryCount();
FeedEvictor.main(
new String[] {
"-feedBasePath", dataPath,
"-retentionType", "instance",
"-retentionLimit", "days(10)",
"-timeZone", "UTC",
"-frequency", "daily",
"-logFile", logFile,
"-falconFeedStorageType", Storage.TYPE.FILESYSTEM.name(),
});
compare(map.get("feed1"), stream.getBuffer());
String expectedInstancePaths = getExpectedInstancePaths(dataPath.replaceAll(storageUrl, ""));
Assert.assertEquals(readLogFile(new Path(logFile)), expectedInstancePaths);
String deletedPath = expectedInstancePaths.split(",")[0].split("=")[1];
Assert.assertFalse(fs.exists(new Path(deletedPath)));
// empty parents
Assert.assertFalse(fs.exists(new Path(deletedPath).getParent()));
Assert.assertFalse(fs.exists(new Path(deletedPath).getParent().getParent()));
// base path not deleted
Assert.assertTrue(fs.exists(new Path("/data/YYYY/feed1/mmHH/dd/MM/")));
// non-eligible empty dirs
long afterDelCount =
fs.getContentSummary(new Path(("/data/YYYY/feed1/mmHH/dd/MM/"))).getDirectoryCount();
Assert.assertEquals((beforeDelCount - afterDelCount), 19);
for (String path : pair.second) {
Assert.assertTrue(fs.exists(new Path(path)));
}
} catch (Exception e) {
Assert.fail("Unknown exception", e);
}
}
@Test
public void testEviction2() throws Exception {
try {
Configuration conf = cluster.getConf();
FileSystem fs = FileSystem.get(conf);
fs.delete(new Path("/"), true);
stream.clear();
Pair<List<String>, List<String>> pair =
createTestData("feed1", "yyyy-MM-dd/'more'/yyyy", 10, TimeUnit.DAYS, "/data");
final String storageUrl = cluster.getConf().get(HadoopClientFactory.FS_DEFAULT_NAME_KEY);
String dataPath =
LocationType.DATA.name()
+ "="
+ storageUrl
+ "/data/YYYY/feed1/mmHH/dd/MM/?{YEAR}-?{MONTH}-?{DAY}/more/?{YEAR}";
String logFile = hdfsUrl + "/falcon/staging/feed/instancePaths-2012-01-01-01-00.csv";
FeedEvictor.main(
new String[] {
"-feedBasePath", dataPath,
"-retentionType", "instance",
"-retentionLimit", "days(10)",
"-timeZone", "UTC",
"-frequency", "daily",
"-logFile", logFile,
"-falconFeedStorageType", Storage.TYPE.FILESYSTEM.name(),
});
assertFailures(fs, pair);
compare(map.get("feed1"), stream.getBuffer());
String expectedInstancePaths = getExpectedInstancePaths(dataPath);
Assert.assertEquals(readLogFile(new Path(logFile)), expectedInstancePaths);
String deletedPath = expectedInstancePaths.split(",")[0].split("=")[1];
Assert.assertFalse(fs.exists(new Path(deletedPath)));
// empty parents
Assert.assertFalse(fs.exists(new Path(deletedPath).getParent()));
Assert.assertFalse(fs.exists(new Path(deletedPath).getParent().getParent()));
// base path not deleted
Assert.assertTrue(fs.exists(new Path("/data/YYYY/feed1/mmHH/dd/MM/")));
} catch (Exception e) {
Assert.fail("Unknown exception", e);
}
}
@Test
public void testEviction3() throws Exception {
try {
Configuration conf = cluster.getConf();
FileSystem fs = FileSystem.get(conf);
fs.delete(new Path("/"), true);
stream.clear();
Pair<List<String>, List<String>> pair =
createTestData("feed2", "yyyyMMddHH/'more'/yyyy", 5, TimeUnit.HOURS, "/data");
final String storageUrl = cluster.getConf().get(HadoopClientFactory.FS_DEFAULT_NAME_KEY);
String dataPath =
LocationType.DATA.name()
+ "="
+ storageUrl
+ "/data/YYYY/feed2/mmHH/dd/MM/?{YEAR}?{MONTH}?{DAY}?{HOUR}/more/?{YEAR}";
String logFile = hdfsUrl + "/falcon/staging/feed/instancePaths-2012-01-01-02-00.csv";
FeedEvictor.main(
new String[] {
"-feedBasePath", dataPath,
"-retentionType", "instance",
"-retentionLimit", "hours(5)",
"-timeZone", "UTC",
"-frequency", "hourly",
"-logFile", logFile,
"-falconFeedStorageType", Storage.TYPE.FILESYSTEM.name(),
});
assertFailures(fs, pair);
compare(map.get("feed2"), stream.getBuffer());
Assert.assertEquals(readLogFile(new Path(logFile)), getExpectedInstancePaths(dataPath));
} catch (Exception e) {
Assert.fail("Unknown exception", e);
}
}