big-data/hadoop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

HDFS-16484. [SPS]: Fix an infinite loop bug in SPSPathIdProcessor thread (#4032)

Browse Source
This commit is contained in:
qinyuren 2022-04-13 10:27:37 +08:00 committed by GitHub
parent e89fd9645b
commit 45394433a1
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 68 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/sps/BlockStorageMovementNeeded.java
View File

@ -17,6 +17,7 @@
*/
package org.apache.hadoop.hdfs.server.namenode.sps;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.util.HashMap;
import java.util.LinkedList;
@ -227,15 +228,18 @@ public synchronized void clearQueuesWithNotification() {
* ID's to process for satisfy the policy.
*/
private class SPSPathIdProcessor implements Runnable {
private static final int MAX_RETRY_COUNT = 3;
@Override
public void run() {
LOG.info("Starting SPSPathIdProcessor!.");
Long startINode = null;
int retryCount = 0;
while (ctxt.isRunning()) {
try {
if (!ctxt.isInSafeMode()) {
if (startINode == null) {
retryCount = 0;
startINode = ctxt.getNextSPSPath();
} // else same id will be retried
if (startINode == null) {
@ -248,7 +252,12 @@ public void run() {
pendingWorkForDirectory.get(startINode);
if (dirPendingWorkInfo != null
&& dirPendingWorkInfo.isDirWorkDone()) {
ctxt.removeSPSHint(startINode);
try {
ctxt.removeSPSHint(startINode);
} catch (FileNotFoundException e) {
// ignore if the file doesn't already exist
startINode = null;
}
pendingWorkForDirectory.remove(startINode);
}
}
@ -268,6 +277,11 @@ public void run() {
LOG.info("Interrupted while waiting in SPSPathIdProcessor", t);
break;
}
retryCount++;
if (retryCount >= MAX_RETRY_COUNT) {
LOG.warn("Skipping this inode {} due to too many retries.", startINode);
startINode = null;
}
}
}
}

53
hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/sps/TestExternalStoragePolicySatisfier.java
View File

@ -201,6 +201,20 @@ private void shutdownCluster() {
}
}
private void stopExternalSps() {
if (externalSps != null) {
externalSps.stopGracefully();
}
}
private void startExternalSps() {
externalSps = new StoragePolicySatisfier(getConf());
externalCtxt = new ExternalSPSContext(externalSps, nnc);
externalSps.init(externalCtxt);
externalSps.start(StoragePolicySatisfierMode.EXTERNAL);
}
private void createCluster() throws IOException {
getConf().setLong("dfs.block.size", DEFAULT_BLOCK_SIZE);
setCluster(startCluster(getConf(), allDiskTypes, NUM_OF_DATANODES,
@ -1445,6 +1459,45 @@ public void testSPSWithDirectoryTreeWithoutFile() throws Exception {
}
}
/**
* Test SPS that satisfy the files and then delete the files before start SPS.
*/
@Test(timeout = 300000)
public void testSPSSatisfyAndThenDeleteFileBeforeStartSPS() throws Exception {
try {
createCluster();
HdfsAdmin hdfsAdmin =
new HdfsAdmin(FileSystem.getDefaultUri(config), config);
StorageType[][] newtypes =
new StorageType[][]{{StorageType.DISK, StorageType.ARCHIVE},
{StorageType.DISK, StorageType.ARCHIVE},
{StorageType.DISK, StorageType.ARCHIVE}};
startAdditionalDNs(config, 3, NUM_OF_DATANODES, newtypes,
STORAGES_PER_DATANODE, CAPACITY, hdfsCluster);
stopExternalSps();
dfs.setStoragePolicy(new Path(FILE), COLD);
hdfsAdmin.satisfyStoragePolicy(new Path(FILE));
dfs.delete(new Path(FILE), true);
startExternalSps();
String file1 = "/testMoveToSatisfyStoragePolicy_1";
writeContent(file1);
dfs.setStoragePolicy(new Path(file1), COLD);
hdfsAdmin.satisfyStoragePolicy(new Path(file1));
hdfsCluster.triggerHeartbeats();
DFSTestUtil.waitExpectedStorageType(file1, StorageType.ARCHIVE, 3, 30000,
dfs);
} finally {
shutdownCluster();
}
}
/**
* Test SPS for directory which has multilevel directories.
*/