private void checkLocalizedPath(boolean visibility)
throws IOException, LoginException, InterruptedException {
TrackerDistributedCacheManager manager =
new TrackerDistributedCacheManager(conf, taskController);
String userName = getJobOwnerName();
File workDir = new File(TEST_ROOT_DIR, "workdir");
Path cacheFile = new Path(TEST_ROOT_DIR, "fourthcachefile");
if (visibility) {
createPublicTempFile(cacheFile);
} else {
createPrivateTempFile(cacheFile);
}
Configuration conf1 = new Configuration(conf);
conf1.set("user.name", userName);
DistributedCache.addCacheFile(cacheFile.toUri(), conf1);
TrackerDistributedCacheManager.determineTimestamps(conf1);
TrackerDistributedCacheManager.determineCacheVisibilities(conf1);
dumpState(conf1);
// Task localizing for job
TaskDistributedCacheManager handle =
manager.newTaskDistributedCacheManager(new JobID("jt", 1), conf1);
handle.setupCache(
conf1,
TaskTracker.getPublicDistributedCacheDir(),
TaskTracker.getPrivateDistributedCacheDir(userName));
JobLocalizer.downloadPrivateCache(conf1);
TaskDistributedCacheManager.CacheFile c = handle.getCacheFiles().get(0);
String distCacheDir;
if (visibility) {
distCacheDir = TaskTracker.getPublicDistributedCacheDir();
} else {
distCacheDir = TaskTracker.getPrivateDistributedCacheDir(userName);
}
Path localizedPath =
manager.getLocalCache(
cacheFile.toUri(),
conf1,
distCacheDir,
fs.getFileStatus(cacheFile),
false,
c.timestamp,
visibility,
c);
assertTrue(
"Cache file didn't get localized in the expected directory. "
+ "Expected localization to happen within "
+ ROOT_MAPRED_LOCAL_DIR
+ "/"
+ distCacheDir
+ ", but was localized at "
+ localizedPath,
localizedPath.toString().contains(distCacheDir));
if (visibility) {
checkPublicFilePermissions(new Path[] {localizedPath});
} else {
checkFilePermissions(new Path[] {localizedPath});
}
}
/**
* This is the typical flow for using the DistributedCache classes.
*
* @throws IOException
* @throws LoginException
*/
public void testManagerFlow() throws IOException, LoginException {
if (!canRun()) {
return;
}
// ****** Imitate JobClient code
// Configures a task/job with both a regular file and a "classpath" file.
Configuration subConf = new Configuration(conf);
String userName = getJobOwnerName();
subConf.set("user.name", userName);
JobID jobid = new JobID("jt", 1);
DistributedCache.addCacheFile(firstCacheFile.toUri(), subConf);
DistributedCache.addFileToClassPath(secondCacheFile, subConf, FileSystem.get(subConf));
TrackerDistributedCacheManager.determineTimestamps(subConf);
TrackerDistributedCacheManager.determineCacheVisibilities(subConf);
// ****** End of imitating JobClient code
Path jobFile = new Path(TEST_ROOT_DIR, "job.xml");
FileOutputStream os = new FileOutputStream(new File(jobFile.toString()));
subConf.writeXml(os);
os.close();
// ****** Imitate TaskRunner code.
TrackerDistributedCacheManager manager =
new TrackerDistributedCacheManager(conf, taskController);
TaskDistributedCacheManager handle = manager.newTaskDistributedCacheManager(jobid, subConf);
assertNull(null, DistributedCache.getLocalCacheFiles(subConf));
File workDir = new File(new Path(TEST_ROOT_DIR, "workdir").toString());
handle.setupCache(
subConf,
TaskTracker.getPublicDistributedCacheDir(),
TaskTracker.getPrivateDistributedCacheDir(userName));
JobLocalizer.downloadPrivateCache(subConf);
// DOESN'T ACTUALLY HAPPEN IN THE TaskRunner (THIS IS A TODO)
// handle.setupPrivateCache(localDirAllocator, TaskTracker
// .getPrivateDistributedCacheDir(userName));
// // ****** End of imitating TaskRunner code
Path[] localCacheFiles = DistributedCache.getLocalCacheFiles(subConf);
assertNotNull(null, localCacheFiles);
assertEquals(2, localCacheFiles.length);
Path cachedFirstFile = localCacheFiles[0];
Path cachedSecondFile = localCacheFiles[1];
assertFileLengthEquals(firstCacheFile, cachedFirstFile);
assertFalse("Paths should be different.", firstCacheFile.equals(cachedFirstFile));
assertEquals(1, handle.getClassPaths().size());
assertEquals(cachedSecondFile.toString(), handle.getClassPaths().get(0));
checkFilePermissions(localCacheFiles);
// Cleanup
handle.release();
manager.purgeCache();
assertFalse(pathToFile(cachedFirstFile).exists());
}
/** test delete cache */
public void testDeleteCache() throws Exception {
if (!canRun()) {
return;
}
// This test needs mapred.local.dir to be single directory
// instead of four, because it assumes that both
// firstcachefile and secondcachefile will be localized on same directory
// so that second localization triggers deleteCache.
// If mapred.local.dir is four directories, second localization might not
// trigger deleteCache, if it is localized in different directory.
Configuration conf2 = new Configuration(conf);
conf2.set("mapred.local.dir", ROOT_MAPRED_LOCAL_DIR.toString());
conf2.setLong("local.cache.size", LOCAL_CACHE_LIMIT);
refreshConf(conf2);
TrackerDistributedCacheManager manager =
new TrackerDistributedCacheManager(conf2, taskController);
FileSystem localfs = FileSystem.getLocal(conf2);
long now = System.currentTimeMillis();
String userName = getJobOwnerName();
conf2.set("user.name", userName);
// We first test the size limit
FileStatus stat = fs.getFileStatus(firstCacheFilePublic);
CacheFile cfile1 =
new CacheFile(
firstCacheFilePublic.toUri(),
CacheFile.FileType.REGULAR,
true,
stat.getModificationTime(),
true);
Path firstLocalCache =
manager.getLocalCache(
firstCacheFilePublic.toUri(),
conf2,
TaskTracker.getPrivateDistributedCacheDir(userName),
fs.getFileStatus(firstCacheFilePublic),
false,
fs.getFileStatus(firstCacheFilePublic).getModificationTime(),
true,
cfile1);
manager.releaseCache(cfile1.getStatus());
// in above code,localized a file of size 4K and then release the cache
// which will cause the cache be deleted when the limit goes out.
// The below code localize another cache which's designed to
// sweep away the first cache.
stat = fs.getFileStatus(secondCacheFilePublic);
CacheFile cfile2 =
new CacheFile(
secondCacheFilePublic.toUri(),
CacheFile.FileType.REGULAR,
true,
stat.getModificationTime(),
true);
assertTrue(
"DistributedCache currently doesn't have cached file", localfs.exists(firstLocalCache));
Path secondLocalCache =
manager.getLocalCache(
secondCacheFilePublic.toUri(),
conf2,
TaskTracker.getPrivateDistributedCacheDir(userName),
fs.getFileStatus(secondCacheFilePublic),
false,
fs.getFileStatus(secondCacheFilePublic).getModificationTime(),
true,
cfile2);
assertFalse(
"DistributedCache failed deleting old" + " cache when the cache store is full.",
localfs.exists(firstLocalCache));
// find the root directory of distributed caches
Path firstCursor = firstLocalCache;
Path secondCursor = secondLocalCache;
while (!firstCursor.equals(secondCursor)) {
// Debug code, to see what these things look like
System.err.println("cursors: " + firstCursor);
System.err.println(" and " + secondCursor);
firstCursor = firstCursor.getParent();
secondCursor = secondCursor.getParent();
}
System.err.println("The final cursor is " + firstCursor);
System.err.println(
"That directory ends up with "
+ localfs.listStatus(firstCursor).length
+ " subdirectories");
Path cachesBase = firstCursor;
assertFalse(
"DistributedCache did not delete the gensym'ed distcache "
+ "directory names when it deleted the files they contained "
+ "because they collectively exceeded the size limit.",
localfs.listStatus(cachesBase).length > 1);
conf2.setLong("local.cache.size", LOCAL_CACHE_LIMIT * 10);
conf2.setLong("mapreduce.tasktracker.local.cache.numberdirectories", LOCAL_CACHE_SUBDIR_LIMIT);
manager = new TrackerDistributedCacheManager(conf2, taskController);
// Now we test the number of sub directories limit
// Create the temporary cache files to be used in the tests.
Path thirdCacheFile = new Path(TEST_ROOT_DIR, "thirdcachefile");
Path fourthCacheFile = new Path(TEST_ROOT_DIR, "fourthcachefile");
// Adding two more small files, so it triggers the number of sub directory
// limit but does not trigger the file size limit.
createPrivateTempFile(thirdCacheFile);
createPrivateTempFile(fourthCacheFile);
DistributedCache.setCacheFiles(new URI[] {thirdCacheFile.toUri()}, conf2);
TrackerDistributedCacheManager.determineCacheVisibilities(conf2);
TrackerDistributedCacheManager.determineTimestamps(conf2);
stat = fs.getFileStatus(thirdCacheFile);
CacheFile cfile3 =
new CacheFile(
thirdCacheFile.toUri(),
CacheFile.FileType.REGULAR,
false,
stat.getModificationTime(),
true);
Path thirdLocalCache =
manager.getLocalCache(
thirdCacheFile.toUri(),
conf2,
TaskTracker.getPrivateDistributedCacheDir(userName),
fs.getFileStatus(thirdCacheFile),
false,
fs.getFileStatus(thirdCacheFile).getModificationTime(),
false,
cfile3);
DistributedCache.setLocalFiles(conf2, thirdLocalCache.toString());
JobLocalizer.downloadPrivateCache(conf2);
// Release the third cache so that it can be deleted while sweeping
manager.releaseCache(cfile3.getStatus());
// Getting the fourth cache will make the number of sub directories becomes
// 3 which is greater than 2. So the released cache will be deleted.
stat = fs.getFileStatus(fourthCacheFile);
CacheFile cfile4 =
new CacheFile(
fourthCacheFile.toUri(),
CacheFile.FileType.REGULAR,
false,
stat.getModificationTime(),
true);
assertTrue(
"DistributedCache currently doesn't have cached file", localfs.exists(thirdLocalCache));
DistributedCache.setCacheFiles(new URI[] {fourthCacheFile.toUri()}, conf2);
DistributedCache.setLocalFiles(conf2, thirdCacheFile.toUri().toString());
TrackerDistributedCacheManager.determineCacheVisibilities(conf2);
TrackerDistributedCacheManager.determineTimestamps(conf2);
Path fourthLocalCache =
manager.getLocalCache(
fourthCacheFile.toUri(),
conf2,
TaskTracker.getPrivateDistributedCacheDir(userName),
fs.getFileStatus(fourthCacheFile),
false,
fs.getFileStatus(fourthCacheFile).getModificationTime(),
false,
cfile4);
assertFalse(
"DistributedCache failed deleting old"
+ " cache when the cache exceeds the number of sub directories limit.",
localfs.exists(thirdLocalCache));
assertFalse(
"DistributedCache did not delete the gensym'ed distcache "
+ "directory names when it deleted the files they contained "
+ "because there were too many.",
localfs.listStatus(cachesBase).length > LOCAL_CACHE_SUBDIR_LIMIT);
// Clean up the files created in this test
new File(thirdCacheFile.toString()).delete();
new File(fourthCacheFile.toString()).delete();
}
public void testReferenceCount()
throws IOException, LoginException, URISyntaxException, InterruptedException {
if (!canRun()) {
return;
}
TrackerDistributedCacheManager manager = new FakeTrackerDistributedCacheManager(conf);
String userName = getJobOwnerName();
File workDir = new File(new Path(TEST_ROOT_DIR, "workdir").toString());
// Configures a job with a regular file
Job job1 = new Job(conf);
Configuration conf1 = job1.getConfiguration();
conf1.set("user.name", userName);
DistributedCache.addCacheFile(secondCacheFile.toUri(), conf1);
TrackerDistributedCacheManager.determineTimestamps(conf1);
TrackerDistributedCacheManager.determineCacheVisibilities(conf1);
// Task localizing for first job
TaskDistributedCacheManager handle =
manager.newTaskDistributedCacheManager(new JobID("jt", 1), conf1);
handle.setupCache(
conf1,
TaskTracker.getPublicDistributedCacheDir(),
TaskTracker.getPrivateDistributedCacheDir(userName));
JobLocalizer.downloadPrivateCache(conf1);
handle.release();
for (TaskDistributedCacheManager.CacheFile c : handle.getCacheFiles()) {
assertEquals(0, manager.getReferenceCount(c.getStatus()));
}
Path thirdCacheFile = new Path(TEST_ROOT_DIR, "thirdcachefile");
createPrivateTempFile(thirdCacheFile);
// Configures another job with three regular files.
Job job2 = new Job(conf);
Configuration conf2 = job2.getConfiguration();
conf2.set("user.name", userName);
// add a file that would get failed to localize
DistributedCache.addCacheFile(firstCacheFilePublic.toUri(), conf2);
// add a file that is already localized by different job
DistributedCache.addCacheFile(secondCacheFile.toUri(), conf2);
// add a file that is never localized
DistributedCache.addCacheFile(thirdCacheFile.toUri(), conf2);
TrackerDistributedCacheManager.determineTimestamps(conf2);
TrackerDistributedCacheManager.determineCacheVisibilities(conf2);
// Task localizing for second job
// localization for the "firstCacheFile" will fail.
handle = manager.newTaskDistributedCacheManager(new JobID("jt", 2), conf2);
Throwable th = null;
try {
handle.setupCache(
conf2,
TaskTracker.getPublicDistributedCacheDir(),
TaskTracker.getPrivateDistributedCacheDir(userName));
JobLocalizer.downloadPrivateCache(conf2);
} catch (IOException e) {
th = e;
LOG.info("Exception during setup", e);
}
assertNotNull(th);
assertTrue(th.getMessage().contains("fake fail"));
handle.release();
th = null;
for (TaskDistributedCacheManager.CacheFile c : handle.getCacheFiles()) {
try {
int refcount = manager.getReferenceCount(c.getStatus());
LOG.info("checking refcount " + c.uri + " of " + refcount);
assertEquals(0, refcount);
} catch (NullPointerException ie) {
th = ie;
LOG.info("Exception getting reference count for " + c.uri, ie);
}
}
assertNotNull(th);
fs.delete(thirdCacheFile, false);
}