HBASE-23322 [hbck2] Simplification on HBCKSCP scheduling

Signed-off-by: Lijin Bin <binlijin@apache.org>
This commit is contained in:
stack 2019-11-19 15:42:32 -08:00
parent 834ccb4bf6
commit 77b4e8c972
8 changed files with 78 additions and 85 deletions

View File

@ -32,7 +32,6 @@ import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Set;
import java.util.function.Function;
import java.util.stream.Collectors;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
@ -71,7 +70,6 @@ import org.apache.hadoop.hbase.ipc.RpcServerInterface;
import org.apache.hadoop.hbase.ipc.ServerRpcController;
import org.apache.hadoop.hbase.master.assignment.RegionStates;
import org.apache.hadoop.hbase.master.locking.LockProcedure;
import org.apache.hadoop.hbase.master.procedure.HBCKServerCrashProcedure;
import org.apache.hadoop.hbase.master.procedure.MasterProcedureEnv;
import org.apache.hadoop.hbase.master.procedure.MasterProcedureUtil;
import org.apache.hadoop.hbase.master.procedure.MasterProcedureUtil.NonceProcedureRunnable;
@ -2587,35 +2585,18 @@ public class MasterRpcServices extends RSRpcServices
public MasterProtos.ScheduleServerCrashProcedureResponse scheduleServerCrashProcedure(
RpcController controller, MasterProtos.ScheduleServerCrashProcedureRequest request)
throws ServiceException {
List<HBaseProtos.ServerName> serverNames = request.getServerNameList();
List<Long> pids = new ArrayList<>();
try {
for (HBaseProtos.ServerName sn: serverNames) {
ServerName serverName = ProtobufUtil.toServerName(sn);
LOG.info("{} schedule ServerCrashProcedure for {}",
this.master.getClientIdAuditPrefix(), serverName);
if (shouldSubmitSCP(serverName)) {
final boolean containsMetaWALs = containMetaWals(serverName);
long pid = this.master.getServerManager().expireServer(serverName,
new Function<ServerName, Long>() {
@Override
public Long apply(ServerName serverName) {
ProcedureExecutor<MasterProcedureEnv> procExec =
master.getMasterProcedureExecutor();
return procExec.submitProcedure(
new HBCKServerCrashProcedure(procExec.getEnvironment(),
serverName, true, containsMetaWALs));
}
});
pids.add(pid);
} else {
pids.add(Procedure.NO_PROC_ID);
}
for (HBaseProtos.ServerName sn: request.getServerNameList()) {
ServerName serverName = ProtobufUtil.toServerName(sn);
LOG.info("{} schedule ServerCrashProcedure for {}",
this.master.getClientIdAuditPrefix(), serverName);
if (shouldSubmitSCP(serverName)) {
pids.add(this.master.getServerManager().expireServer(serverName, true));
} else {
pids.add(Procedure.NO_PROC_ID);
}
return MasterProtos.ScheduleServerCrashProcedureResponse.newBuilder().addAllPid(pids).build();
} catch (IOException e) {
throw new ServiceException(e);
}
return MasterProtos.ScheduleServerCrashProcedureResponse.newBuilder().addAllPid(pids).build();
}
@Override

View File

@ -33,7 +33,6 @@ import java.util.concurrent.ConcurrentNavigableMap;
import java.util.concurrent.ConcurrentSkipListMap;
import java.util.concurrent.CopyOnWriteArrayList;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.function.Function;
import java.util.function.Predicate;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
@ -562,32 +561,18 @@ public class ServerManager {
/**
* Expire the passed server. Add it to list of dead servers and queue a shutdown processing.
* @return True if we queued a ServerCrashProcedure else false if we did not (could happen for
* many reasons including the fact that its this server that is going down or we already
* have queued an SCP for this server or SCP processing is currently disabled because we
* are in startup phase).
* @return pid if we queued a ServerCrashProcedure else {@link Procedure#NO_PROC_ID} if we did
* not (could happen for many reasons including the fact that its this server that is
* going down or we already have queued an SCP for this server or SCP processing is
* currently disabled because we are in startup phase).
*/
public boolean expireServer(final ServerName serverName) {
return expireServer(serverName, new Function<ServerName, Long>() {
@Override
public Long apply(ServerName serverName) {
return master.getAssignmentManager().submitServerCrash(serverName, true);
}
}) != Procedure.NO_PROC_ID;
@VisibleForTesting // Redo test so we can make this protected.
public synchronized long expireServer(final ServerName serverName) {
return expireServer(serverName, false);
}
/**
* Expire the passed server. Add it to list of dead servers and queue a shutdown processing.
* Used when expireServer is externally invoked by hbck2.
* @param function Takes ServerName and returns pid. See default implementation which queues
* an SCP via the AssignmentManager.
* @return True if we queued a ServerCrashProcedure else false if we did not (could happen for
* many reasons including the fact that its this server that is going down or we already
* have queued an SCP for this server or SCP processing is currently disabled because we
* are in startup phase).
*/
synchronized long expireServer(final ServerName serverName,
Function<ServerName, Long> function) {
synchronized long expireServer(final ServerName serverName, boolean force) {
// THIS server is going down... can't handle our own expiration.
if (serverName.equals(master.getServerName())) {
if (!(master.isAborted() || master.isStopped())) {
@ -612,10 +597,7 @@ public class ServerManager {
return Procedure.NO_PROC_ID;
}
LOG.info("Processing expiration of " + serverName + " on " + this.master.getServerName());
long pid = function.apply(serverName);
if (pid <= 0) {
return Procedure.NO_PROC_ID;
}
long pid = master.getAssignmentManager().submitServerCrash(serverName, true, force);
// Tell our listeners that a server was removed
if (!this.listeners.isEmpty()) {
this.listeners.stream().forEach(l -> l.serverRemoved(serverName));

View File

@ -58,6 +58,7 @@ import org.apache.hadoop.hbase.master.RegionState.State;
import org.apache.hadoop.hbase.master.ServerManager;
import org.apache.hadoop.hbase.master.TableStateManager;
import org.apache.hadoop.hbase.master.balancer.FavoredStochasticBalancer;
import org.apache.hadoop.hbase.master.procedure.HBCKServerCrashProcedure;
import org.apache.hadoop.hbase.master.procedure.MasterProcedureEnv;
import org.apache.hadoop.hbase.master.procedure.MasterProcedureScheduler;
import org.apache.hadoop.hbase.master.procedure.ProcedureSyncWait;
@ -1490,15 +1491,21 @@ public class AssignmentManager {
return 0;
}
public long submitServerCrash(ServerName serverName, boolean shouldSplitWal) {
boolean carryingMeta;
long pid;
/**
* Usually run by the Master in reaction to server crash during normal processing.
* Can also be invoked via external RPC to effect repair; in the latter case,
* the 'force' flag is set so we push through the SCP though context may indicate
* already-running-SCP (An old SCP may have exited abnormally, or damaged cluster
* may still have references in hbase:meta to 'Unknown Servers' -- servers that
* are not online or in dead servers list, etc.)
* @param force Set if the request came in externally over RPC (via hbck2). Force means
* run the SCP even if it seems as though there might be an outstanding
* SCP running.
* @return pid of scheduled SCP or {@link Procedure#NO_PROC_ID} if none scheduled.
*/
public long submitServerCrash(ServerName serverName, boolean shouldSplitWal, boolean force) {
// May be an 'Unknown Server' so handle case where serverNode is null.
ServerStateNode serverNode = regionStates.getServerNode(serverName);
if (serverNode == null) {
LOG.info("Skip to add SCP for {} since this server should be OFFLINE already", serverName);
return -1;
}
// Remove the in-memory rsReports result
synchronized (rsReports) {
rsReports.remove(serverName);
@ -1508,26 +1515,43 @@ public class AssignmentManager {
// server state to CRASHED, we will no longer accept the reportRegionStateTransition call from
// this server. This is used to simplify the implementation for TRSP and SCP, where we can make
// sure that, the region list fetched by SCP will not be changed any more.
serverNode.writeLock().lock();
if (serverNode != null) {
serverNode.writeLock().lock();
}
boolean carryingMeta;
long pid;
try {
ProcedureExecutor<MasterProcedureEnv> procExec = this.master.getMasterProcedureExecutor();
carryingMeta = isCarryingMeta(serverName);
if (!serverNode.isInState(ServerState.ONLINE)) {
LOG.info(
"Skip to add SCP for {} with meta= {}, " +
"since there should be a SCP is processing or already done for this server node",
serverName, carryingMeta);
return -1;
if (!force && serverNode != null && !serverNode.isInState(ServerState.ONLINE)) {
LOG.info("Skip adding SCP for {} (meta={}) -- running?", serverNode, carryingMeta);
return Procedure.NO_PROC_ID;
} else {
serverNode.setState(ServerState.CRASHED);
pid = procExec.submitProcedure(new ServerCrashProcedure(procExec.getEnvironment(),
serverName, shouldSplitWal, carryingMeta));
LOG.info(
"Added {} to dead servers which carryingMeta={}, submitted ServerCrashProcedure pid={}",
serverName, carryingMeta, pid);
MasterProcedureEnv mpe = procExec.getEnvironment();
// If serverNode == null, then 'Unknown Server'. Schedule HBCKSCP instead.
// HBCKSCP scours Master in-memory state AND hbase;meta for references to
// serverName just-in-case. An SCP that is scheduled when the server is
// 'Unknown' probably originated externally with HBCK2 fix-it tool.
ServerState oldState = null;
if (serverNode != null) {
oldState = serverNode.getState();
serverNode.setState(ServerState.CRASHED);
}
if (force) {
pid = procExec.submitProcedure(
new HBCKServerCrashProcedure(mpe, serverName, shouldSplitWal, carryingMeta));
} else {
pid = procExec.submitProcedure(
new ServerCrashProcedure(mpe, serverName, shouldSplitWal, carryingMeta));
}
LOG.info("Scheduled SCP pid={} for {} (carryingMeta={}){}.", pid, serverName, carryingMeta,
serverNode == null? "": " " + serverNode.toString() + ", oldState=" + oldState);
}
} finally {
serverNode.writeLock().unlock();
if (serverNode != null) {
serverNode.writeLock().unlock();
}
}
return pid;
}

View File

@ -714,6 +714,9 @@ public class RegionStates {
serverMap.remove(serverName);
}
/**
* @return Pertinent ServerStateNode or NULL if none found.
*/
@VisibleForTesting
public ServerStateNode getServerNode(final ServerName serverName) {
return serverMap.get(serverName);

View File

@ -33,12 +33,9 @@ import org.apache.yetus.audience.InterfaceAudience;
*/
@InterfaceAudience.Private
public class ServerStateNode implements Comparable<ServerStateNode> {
private final Set<RegionStateNode> regions;
private final ServerName serverName;
private final ReadWriteLock lock = new ReentrantReadWriteLock();
private volatile ServerState state = ServerState.ONLINE;
public ServerStateNode(ServerName serverName) {
@ -120,6 +117,7 @@ public class ServerStateNode implements Comparable<ServerStateNode> {
@Override
public String toString() {
return String.format("ServerStateNode(%s)", getServerName());
return getServerName() + "/" + getState() + "/regionCount=" + this.regions.size() +
"/lock=" + this.lock;
}
}

View File

@ -81,12 +81,15 @@ public class HBCKServerCrashProcedure extends ServerCrashProcedure {
LOG.warn("Failed get of all regions; continuing", ioe);
}
if (ps == null || ps.isEmpty()) {
LOG.warn("No regions found in hbase:meta");
return ris;
}
List<RegionInfo> aggregate = ris == null || ris.isEmpty()?
new ArrayList<>(): new ArrayList<>(ris);
int before = aggregate.size();
ps.stream().filter(p -> p.getSecond() != null && p.getSecond().equals(getServerName())).
forEach(p -> aggregate.add(p.getFirst()));
LOG.info("Found {} mentions of {} in hbase:meta", aggregate.size() - before, getServerName());
return aggregate;
}
}

View File

@ -109,7 +109,8 @@ public class TestClusterRestartFailover extends AbstractTestRestartCluster {
((ServerCrashProcedure) p).getServerName().equals(SERVER_FOR_TEST)).findAny();
assertTrue("Should have one SCP for " + SERVER_FOR_TEST, procedure.isPresent());
assertFalse("Submit the SCP for the same serverName " + SERVER_FOR_TEST + " which should fail",
UTIL.getHBaseCluster().getMaster().getServerManager().expireServer(SERVER_FOR_TEST));
UTIL.getHBaseCluster().getMaster().getServerManager().expireServer(SERVER_FOR_TEST) ==
Procedure.NO_PROC_ID);
// Wait the SCP to finish
SCP_LATCH.countDown();
@ -117,7 +118,8 @@ public class TestClusterRestartFailover extends AbstractTestRestartCluster {
assertFalse("Even when the SCP is finished, the duplicate SCP should not be scheduled for " +
SERVER_FOR_TEST,
UTIL.getHBaseCluster().getMaster().getServerManager().expireServer(SERVER_FOR_TEST));
UTIL.getHBaseCluster().getMaster().getServerManager().expireServer(SERVER_FOR_TEST) ==
Procedure.NO_PROC_ID);
serverNode = UTIL.getHBaseCluster().getMaster().getAssignmentManager().getRegionStates()
.getServerNode(SERVER_FOR_TEST);
assertNull("serverNode should be deleted after SCP finished", serverNode);

View File

@ -304,7 +304,7 @@ public abstract class TestAssignmentManagerBase {
protected void doCrash(final ServerName serverName) {
this.master.getServerManager().moveFromOnlineToDeadServers(serverName);
this.am.submitServerCrash(serverName, false/* No WALs here */);
this.am.submitServerCrash(serverName, false/* No WALs here */, false);
// add a new server to avoid killing all the region servers which may hang the UTs
ServerName newSn = ServerName.valueOf("localhost", 10000 + newRsAdded, 1);
newRsAdded++;