HBASE-21072 Block out HBCK1 in hbase2

Write the hbase-1.x hbck1 lock file to block out hbck1 instances writing
state to an hbase-2.x cluster (could do damage).
Set hbase.write.hbck1.lock.file to false to disable this writing.
This commit is contained in:
Michael Stack 2018-08-22 22:44:00 -07:00
parent 8a9acd4d2a
commit 66add55234
3 changed files with 128 additions and 35 deletions

View File

@ -185,6 +185,7 @@ import org.apache.hadoop.hbase.util.Addressing;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.CompressionTest;
import org.apache.hadoop.hbase.util.EncryptionTest;
import org.apache.hadoop.hbase.util.HBaseFsck;
import org.apache.hadoop.hbase.util.HFileArchiveUtil;
import org.apache.hadoop.hbase.util.HasThread;
import org.apache.hadoop.hbase.util.IdLock;
@ -864,7 +865,13 @@ public class HMaster extends HRegionServer implements MasterServices {
ZKClusterId.setClusterId(this.zooKeeper, fileSystemManager.getClusterId());
this.clusterId = clusterId.toString();
// Precaution. Put in place the old hbck1 lock file to fence out old hbase1s running their
// hbck1s against an hbase2 cluster; it could do damage. To skip this behavior, set
// hbase.write.hbck1.lock.file to false.
if (this.conf.getBoolean("hbase.write.hbck1.lock.file", true)) {
HBaseFsck.checkAndMarkRunningHbck(this.conf,
HBaseFsck.createLockRetryCounterFactory(this.conf).create());
}
status.setStatus("Initialze ServerManager and schedule SCP for crash servers");
this.serverManager = createServerManager(this);

View File

@ -144,6 +144,7 @@ import org.apache.hadoop.security.UserGroupInformation;
import org.apache.hadoop.util.ReflectionUtils;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner;
import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;
import org.apache.hbase.thirdparty.com.google.common.collect.Sets;
import org.apache.yetus.audience.InterfaceAudience;
import org.apache.yetus.audience.InterfaceStability;
@ -164,7 +165,10 @@ import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.AdminServic
/**
* HBaseFsck (hbck) is a tool for checking and repairing region consistency and
* table integrity problems in a corrupted HBase.
* table integrity problems in a corrupted HBase. This tool was written for hbase-1.x. It does not
* work with hbase-2.x; it can read state but is not allowed to change state; i.e. effect 'repair'.
* See hbck2 (HBASE-19121) for a hbck tool for hbase2.
*
* <p>
* Region consistency checks verify that hbase:meta, region deployment on region
* servers and the state of data in HDFS (.regioninfo files) all are in
@ -217,7 +221,12 @@ public class HBaseFsck extends Configured implements Closeable {
private static final int DEFAULT_OVERLAPS_TO_SIDELINE = 2;
private static final int DEFAULT_MAX_MERGE = 5;
private static final String TO_BE_LOADED = "to_be_loaded";
private static final String HBCK_LOCK_FILE = "hbase-hbck.lock";
/**
* Here is where hbase-1.x used to default the lock for hbck1.
* It puts in place a lock when it goes to write/make changes.
*/
@VisibleForTesting
public static final String HBCK_LOCK_FILE = "hbase-hbck.lock";
private static final int DEFAULT_MAX_LOCK_FILE_ATTEMPTS = 5;
private static final int DEFAULT_LOCK_FILE_ATTEMPT_SLEEP_INTERVAL = 200; // milliseconds
private static final int DEFAULT_LOCK_FILE_ATTEMPT_MAX_SLEEP_TIME = 5000; // milliseconds
@ -369,40 +378,75 @@ public class HBaseFsck extends Configured implements Closeable {
super(conf);
errors = getErrorReporter(getConf());
this.executor = exec;
lockFileRetryCounterFactory = new RetryCounterFactory(
getConf().getInt("hbase.hbck.lockfile.attempts", DEFAULT_MAX_LOCK_FILE_ATTEMPTS),
getConf().getInt(
"hbase.hbck.lockfile.attempt.sleep.interval", DEFAULT_LOCK_FILE_ATTEMPT_SLEEP_INTERVAL),
getConf().getInt(
"hbase.hbck.lockfile.attempt.maxsleeptime", DEFAULT_LOCK_FILE_ATTEMPT_MAX_SLEEP_TIME));
createZNodeRetryCounterFactory = new RetryCounterFactory(
getConf().getInt("hbase.hbck.createznode.attempts", DEFAULT_MAX_CREATE_ZNODE_ATTEMPTS),
getConf().getInt(
"hbase.hbck.createznode.attempt.sleep.interval",
DEFAULT_CREATE_ZNODE_ATTEMPT_SLEEP_INTERVAL),
getConf().getInt(
"hbase.hbck.createznode.attempt.maxsleeptime",
DEFAULT_CREATE_ZNODE_ATTEMPT_MAX_SLEEP_TIME));
lockFileRetryCounterFactory = createLockRetryCounterFactory(getConf());
createZNodeRetryCounterFactory = createZnodeRetryCounterFactory(getConf());
zkw = createZooKeeperWatcher();
}
private class FileLockCallable implements Callable<FSDataOutputStream> {
RetryCounter retryCounter;
/**
* @return A retry counter factory configured for retrying lock file creation.
*/
public static RetryCounterFactory createLockRetryCounterFactory(Configuration conf) {
return new RetryCounterFactory(
conf.getInt("hbase.hbck.lockfile.attempts", DEFAULT_MAX_LOCK_FILE_ATTEMPTS),
conf.getInt("hbase.hbck.lockfile.attempt.sleep.interval",
DEFAULT_LOCK_FILE_ATTEMPT_SLEEP_INTERVAL),
conf.getInt("hbase.hbck.lockfile.attempt.maxsleeptime",
DEFAULT_LOCK_FILE_ATTEMPT_MAX_SLEEP_TIME));
}
public FileLockCallable(RetryCounter retryCounter) {
/**
* @return A retry counter factory configured for retrying znode creation.
*/
private static RetryCounterFactory createZnodeRetryCounterFactory(Configuration conf) {
return new RetryCounterFactory(
conf.getInt("hbase.hbck.createznode.attempts", DEFAULT_MAX_CREATE_ZNODE_ATTEMPTS),
conf.getInt("hbase.hbck.createznode.attempt.sleep.interval",
DEFAULT_CREATE_ZNODE_ATTEMPT_SLEEP_INTERVAL),
conf.getInt("hbase.hbck.createznode.attempt.maxsleeptime",
DEFAULT_CREATE_ZNODE_ATTEMPT_MAX_SLEEP_TIME));
}
/**
* @return Return the tmp dir this tool writes too.
*/
@VisibleForTesting
public static Path getTmpDir(Configuration conf) throws IOException {
return new Path(FSUtils.getRootDir(conf), HConstants.HBASE_TEMP_DIRECTORY);
}
private static class FileLockCallable implements Callable<FSDataOutputStream> {
RetryCounter retryCounter;
private final Configuration conf;
private Path hbckLockPath = null;
public FileLockCallable(Configuration conf, RetryCounter retryCounter) {
this.retryCounter = retryCounter;
this.conf = conf;
}
/**
* @return Will be <code>null</code> unless you call {@link #call()}
*/
Path getHbckLockPath() {
return this.hbckLockPath;
}
@Override
public FSDataOutputStream call() throws IOException {
try {
FileSystem fs = FSUtils.getCurrentFileSystem(getConf());
FsPermission defaultPerms = FSUtils.getFilePermissions(fs, getConf(),
FileSystem fs = FSUtils.getCurrentFileSystem(this.conf);
FsPermission defaultPerms = FSUtils.getFilePermissions(fs, this.conf,
HConstants.DATA_FILE_UMASK_KEY);
Path tmpDir = new Path(FSUtils.getRootDir(getConf()), HConstants.HBASE_TEMP_DIRECTORY);
Path tmpDir = getTmpDir(conf);
this.hbckLockPath = new Path(tmpDir, HBCK_LOCK_FILE);
fs.mkdirs(tmpDir);
HBCK_LOCK_PATH = new Path(tmpDir, HBCK_LOCK_FILE);
final FSDataOutputStream out = createFileWithRetries(fs, HBCK_LOCK_PATH, defaultPerms);
final FSDataOutputStream out = createFileWithRetries(fs, this.hbckLockPath, defaultPerms);
out.writeBytes(InetAddress.getLocalHost().toString());
// Add a note into the file we write on why hbase2 is writing out an hbck1 lock file.
out.writeBytes(" Written by an hbase-2.x Master to block an " +
"attempt by an hbase-1.x HBCK tool making modification to state. " +
"See 'HBCK must match HBase server version' in the hbase refguide.");
out.flush();
return out;
} catch(RemoteException e) {
@ -417,7 +461,6 @@ public class HBaseFsck extends Configured implements Closeable {
private FSDataOutputStream createFileWithRetries(final FileSystem fs,
final Path hbckLockFilePath, final FsPermission defaultPerms)
throws IOException {
IOException exception = null;
do {
try {
@ -449,13 +492,13 @@ public class HBaseFsck extends Configured implements Closeable {
* @return FSDataOutputStream object corresponding to the newly opened lock file
* @throws IOException if IO failure occurs
*/
private FSDataOutputStream checkAndMarkRunningHbck() throws IOException {
RetryCounter retryCounter = lockFileRetryCounterFactory.create();
FileLockCallable callable = new FileLockCallable(retryCounter);
public static Pair<Path, FSDataOutputStream> checkAndMarkRunningHbck(Configuration conf,
RetryCounter retryCounter) throws IOException {
FileLockCallable callable = new FileLockCallable(conf, retryCounter);
ExecutorService executor = Executors.newFixedThreadPool(1);
FutureTask<FSDataOutputStream> futureTask = new FutureTask<>(callable);
executor.execute(futureTask);
final int timeoutInSeconds = getConf().getInt(
final int timeoutInSeconds = conf.getInt(
"hbase.hbck.lockfile.maxwaittime", DEFAULT_WAIT_FOR_LOCK_TIMEOUT);
FSDataOutputStream stream = null;
try {
@ -472,7 +515,7 @@ public class HBaseFsck extends Configured implements Closeable {
} finally {
executor.shutdownNow();
}
return stream;
return new Pair<Path, FSDataOutputStream>(callable.getHbckLockPath(), stream);
}
private void unlockHbck() {
@ -481,8 +524,7 @@ public class HBaseFsck extends Configured implements Closeable {
do {
try {
IOUtils.closeQuietly(hbckOutFd);
FSUtils.delete(FSUtils.getCurrentFileSystem(getConf()),
HBCK_LOCK_PATH, true);
FSUtils.delete(FSUtils.getCurrentFileSystem(getConf()), HBCK_LOCK_PATH, true);
LOG.info("Finishing hbck");
return;
} catch (IOException ioe) {
@ -511,7 +553,10 @@ public class HBaseFsck extends Configured implements Closeable {
if (isExclusive()) {
// Grab the lock
hbckOutFd = checkAndMarkRunningHbck();
Pair<Path, FSDataOutputStream> pair =
checkAndMarkRunningHbck(getConf(), this.lockFileRetryCounterFactory.create());
HBCK_LOCK_PATH = pair.getFirst();
this.hbckOutFd = pair.getSecond();
if (hbckOutFd == null) {
setRetCode(-1);
LOG.error("Another instance of hbck is fixing HBase, exiting this instance. " +

View File

@ -19,11 +19,15 @@ package org.apache.hadoop.hbase.master;
import static org.junit.Assert.assertArrayEquals;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertTrue;
import static org.junit.Assert.fail;
import java.io.IOException;
import java.util.List;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.HBaseClassTestRule;
import org.apache.hadoop.hbase.HBaseTestingUtility;
import org.apache.hadoop.hbase.HColumnDescriptor;
@ -40,10 +44,10 @@ import org.apache.hadoop.hbase.client.RegionInfo;
import org.apache.hadoop.hbase.client.RegionInfoBuilder;
import org.apache.hadoop.hbase.client.Table;
import org.apache.hadoop.hbase.client.TableState;
import org.apache.hadoop.hbase.master.assignment.RegionStates;
import org.apache.hadoop.hbase.testclassification.MasterTests;
import org.apache.hadoop.hbase.testclassification.MediumTests;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.HBaseFsck;
import org.apache.hadoop.hbase.util.Pair;
import org.apache.hadoop.util.StringUtils;
import org.junit.AfterClass;
@ -192,5 +196,42 @@ public class TestMaster {
TEST_UTIL.deleteTable(tableName);
}
}
@Test
public void testBlockingHbkc1WithLockFile() throws IOException {
// This is how the patch to the lock file is created inside in HBaseFsck. Too hard to use its
// actual method without disturbing HBaseFsck... Do the below mimic instead.
Path hbckLockPath = new Path(HBaseFsck.getTmpDir(TEST_UTIL.getConfiguration()),
HBaseFsck.HBCK_LOCK_FILE);
FileSystem fs = TEST_UTIL.getTestFileSystem();
assertTrue(fs.exists(hbckLockPath));
TEST_UTIL.getMiniHBaseCluster().
killMaster(TEST_UTIL.getMiniHBaseCluster().getMaster().getServerName());
assertTrue(fs.exists(hbckLockPath));
TEST_UTIL.getMiniHBaseCluster().startMaster();
TEST_UTIL.waitFor(30000, () -> TEST_UTIL.getMiniHBaseCluster().getMaster() != null &&
TEST_UTIL.getMiniHBaseCluster().getMaster().isInitialized());
assertTrue(fs.exists(hbckLockPath));
// Start a second Master. Should be fine.
TEST_UTIL.getMiniHBaseCluster().startMaster();
assertTrue(fs.exists(hbckLockPath));
fs.delete(hbckLockPath, true);
assertFalse(fs.exists(hbckLockPath));
// Kill all Masters.
TEST_UTIL.getMiniHBaseCluster().getLiveMasterThreads().stream().
map(sn -> sn.getMaster().getServerName()).forEach(sn -> {
try {
TEST_UTIL.getMiniHBaseCluster().killMaster(sn);
} catch (IOException e) {
e.printStackTrace();
}
});
// Start a new one.
TEST_UTIL.getMiniHBaseCluster().startMaster();
TEST_UTIL.waitFor(30000, () -> TEST_UTIL.getMiniHBaseCluster().getMaster() != null &&
TEST_UTIL.getMiniHBaseCluster().getMaster().isInitialized());
// Assert lock gets put in place again.
assertTrue(fs.exists(hbckLockPath));
}
}