/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.hbase.master.balancer;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.NavigableMap;
import java.util.Random;
import java.util.TreeMap;
import java.util.Comparator;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.hbase.HConstants;
import org.apache.hadoop.hbase.classification.InterfaceAudience;
import org.apache.hadoop.hbase.HBaseIOException;
import org.apache.hadoop.hbase.HBaseInterfaceAudience;
import org.apache.hadoop.hbase.HRegionInfo;
import org.apache.hadoop.hbase.ServerName;
import org.apache.hadoop.hbase.TableName;
import org.apache.hadoop.hbase.master.RegionPlan;
import com.google.common.collect.MinMaxPriorityQueue;
import org.apache.hadoop.hbase.util.Pair;
/**
* Makes decisions about the placement and movement of Regions across
* RegionServers.
*
* <p>Cluster-wide load balancing will occur only when there are no regions in
* transition and according to a fixed period of a time using {@link #balanceCluster(Map)}.
*
* <p>On cluster startup, bulk assignment can be used to determine
* locations for all Regions in a cluster.
*
* <p>This classes produces plans for the
* {@link org.apache.hadoop.hbase.master.AssignmentManager} to execute.
*/
@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.CONFIG)
public class SimpleLoadBalancer extends BaseLoadBalancer {
private static final Log LOG = LogFactory.getLog(SimpleLoadBalancer.class);
private static final Random RANDOM = new Random(System.currentTimeMillis());
private RegionInfoComparator riComparator = new RegionInfoComparator();
private RegionPlan.RegionPlanComparator rpComparator = new RegionPlan.RegionPlanComparator();
private float avgLoadOverall;
private List<ServerAndLoad> serverLoadList;
/**
* Stores additional per-server information about the regions added/removed
* during the run of the balancing algorithm.
*
* For servers that shed regions, we need to track which regions we have already
* shed. <b>nextRegionForUnload</b> contains the index in the list of regions on
* the server that is the next to be shed.
*/
static class BalanceInfo {
private int nextRegionForUnload;
private int numRegionsAdded;
private List<HRegionInfo> hriList;
public BalanceInfo(int nextRegionForUnload, int numRegionsAdded, List<HRegionInfo> hriList) {
this.nextRegionForUnload = nextRegionForUnload;
this.numRegionsAdded = numRegionsAdded;
this.hriList = hriList;
}
int getNextRegionForUnload() {
return nextRegionForUnload;
}
int getNumRegionsAdded() {
return numRegionsAdded;
}
void setNumRegionsAdded(int numAdded) {
this.numRegionsAdded = numAdded;
}
List<HRegionInfo> getHriList() {
return hriList;
}
void setNextRegionForUnload(int nextRegionForUnload) {
this.nextRegionForUnload = nextRegionForUnload;
}
}
public void setClusterLoad(Map<TableName, Map<ServerName, List<HRegionInfo>>> clusterLoad){
serverLoadList = new ArrayList<>();
float sum = 0;
for(Map.Entry<TableName, Map<ServerName, List<HRegionInfo>>> clusterEntry : clusterLoad.entrySet()){
for(Map.Entry<ServerName, List<HRegionInfo>> entry : clusterEntry.getValue().entrySet()){
if(entry.getKey().equals(masterServerName)) continue; // we shouldn't include master as potential assignee
serverLoadList.add(new ServerAndLoad(entry.getKey(), entry.getValue().size()));
sum += entry.getValue().size();
}
}
avgLoadOverall = sum / serverLoadList.size();
}
@Override
public void onConfigurationChange(Configuration conf) {
float originSlop = slop;
float originOverallSlop = overallSlop;
super.setConf(conf);
LOG.info("Update configuration of SimpleLoadBalancer, previous slop is "
+ originSlop + ", current slop is " + slop + "previous overallSlop is" +
originOverallSlop + ", current overallSlop is " + originOverallSlop);
}
private void setLoad(List<ServerAndLoad> slList, int i, int loadChange){
ServerAndLoad newsl = new ServerAndLoad(slList.get(i).getServerName(),slList.get(i).getLoad() + loadChange);
slList.set(i, newsl);
}
/**
* A checker function to decide when we want balance overall and certain table has been balanced,
* do we still need to re-distribute regions of this table to achieve the state of overall-balance
* @return true if this table should be balanced.
*/
private boolean overallNeedsBalance() {
int floor = (int) Math.floor(avgLoadOverall * (1 - overallSlop));
int ceiling = (int) Math.ceil(avgLoadOverall * (1 + overallSlop));
int max = 0, min = Integer.MAX_VALUE;
for(ServerAndLoad server : serverLoadList){
max = Math.max(server.getLoad(), max);
min = Math.min(server.getLoad(), min);
}
if (max <= ceiling && min >= floor) {
if (LOG.isTraceEnabled()) {
// If nothing to balance, then don't say anything unless trace-level logging.
LOG.trace("Skipping load balancing because cluster is balanced at overall level");
}
return false;
}
return true;
}
/**
* Generate a global load balancing plan according to the specified map of
* server information to the most loaded regions of each server.
*
* The load balancing invariant is that all servers are within 1 region of the
* average number of regions per server. If the average is an integer number,
* all servers will be balanced to the average. Otherwise, all servers will
* have either floor(average) or ceiling(average) regions.
*
* HBASE-3609 Modeled regionsToMove using Guava's MinMaxPriorityQueue so that
* we can fetch from both ends of the queue.
* At the beginning, we check whether there was empty region server
* just discovered by Master. If so, we alternately choose new / old
* regions from head / tail of regionsToMove, respectively. This alternation
* avoids clustering young regions on the newly discovered region server.
* Otherwise, we choose new regions from head of regionsToMove.
*
* Another improvement from HBASE-3609 is that we assign regions from
* regionsToMove to underloaded servers in round-robin fashion.
* Previously one underloaded server would be filled before we move onto
* the next underloaded server, leading to clustering of young regions.
*
* Finally, we randomly shuffle underloaded servers so that they receive
* offloaded regions relatively evenly across calls to balanceCluster().
*
* The algorithm is currently implemented as such:
*
* <ol>
* <li>Determine the two valid numbers of regions each server should have,
* <b>MIN</b>=floor(average) and <b>MAX</b>=ceiling(average).
*
* <li>Iterate down the most loaded servers, shedding regions from each so
* each server hosts exactly <b>MAX</b> regions. Stop once you reach a
* server that already has <= <b>MAX</b> regions.
* <p>
* Order the regions to move from most recent to least.
*
* <li>Iterate down the least loaded servers, assigning regions so each server
* has exactly <b>MIN</b> regions. Stop once you reach a server that
* already has >= <b>MIN</b> regions.
*
* Regions being assigned to underloaded servers are those that were shed
* in the previous step. It is possible that there were not enough
* regions shed to fill each underloaded server to <b>MIN</b>. If so we
* end up with a number of regions required to do so, <b>neededRegions</b>.
*
* It is also possible that we were able to fill each underloaded but ended
* up with regions that were unassigned from overloaded servers but that
* still do not have assignment.
*
* If neither of these conditions hold (no regions needed to fill the
* underloaded servers, no regions leftover from overloaded servers),
* we are done and return. Otherwise we handle these cases below.
*
* <li>If <b>neededRegions</b> is non-zero (still have underloaded servers),
* we iterate the most loaded servers again, shedding a single server from
* each (this brings them from having <b>MAX</b> regions to having
* <b>MIN</b> regions).
*
* <li>We now definitely have more regions that need assignment, either from
* the previous step or from the original shedding from overloaded servers.
* Iterate the least loaded servers filling each to <b>MIN</b>.
*
* <li>If we still have more regions that need assignment, again iterate the
* least loaded servers, this time giving each one (filling them to
* <b>MAX</b>) until we run out.
*
* <li>All servers will now either host <b>MIN</b> or <b>MAX</b> regions.
*
* In addition, any server hosting >= <b>MAX</b> regions is guaranteed
* to end up with <b>MAX</b> regions at the end of the balancing. This
* ensures the minimal number of regions possible are moved.
* </ol>
*
* TODO: We can at-most reassign the number of regions away from a particular
* server to be how many they report as most loaded.
* Should we just keep all assignment in memory? Any objections?
* Does this mean we need HeapSize on HMaster? Or just careful monitor?
* (current thinking is we will hold all assignments in memory)
*
* @param clusterMap Map of regionservers and their load/region information to
* a list of their most loaded regions
* @return a list of regions to be moved, including source and destination,
* or null if cluster is already balanced
*/
@Override
public List<RegionPlan> balanceCluster(
Map<ServerName, List<HRegionInfo>> clusterMap) {
List<RegionPlan> regionsToReturn = balanceMasterRegions(clusterMap);
if (regionsToReturn != null || clusterMap == null || clusterMap.size() <= 1) {
return regionsToReturn;
}
if (masterServerName != null && clusterMap.containsKey(masterServerName)) {
if (clusterMap.size() <= 2) {
return null;
}
clusterMap = new HashMap<>(clusterMap);
clusterMap.remove(masterServerName);
}
long startTime = System.currentTimeMillis();
// construct a Cluster object with clusterMap and rest of the
// argument as defaults
Cluster c = new Cluster(clusterMap, null, this.regionFinder, this.rackManager);
if (!this.needsBalance(c) && !this.overallNeedsBalance()) return null;
ClusterLoadState cs = new ClusterLoadState(clusterMap);
int numServers = cs.getNumServers();
NavigableMap<ServerAndLoad, List<HRegionInfo>> serversByLoad = cs.getServersByLoad();
int numRegions = cs.getNumRegions();
float average = cs.getLoadAverage();
int max = (int)Math.ceil(average);
int min = (int)average;
// Using to check balance result.
StringBuilder strBalanceParam = new StringBuilder();
strBalanceParam.append("Balance parameter: numRegions=").append(numRegions)
.append(", numServers=").append(numServers).append(", max=").append(max)
.append(", min=").append(min);
LOG.debug(strBalanceParam.toString());
// Balance the cluster
// TODO: Look at data block locality or a more complex load to do this
MinMaxPriorityQueue<RegionPlan> regionsToMove =
MinMaxPriorityQueue.orderedBy(rpComparator).create();
regionsToReturn = new ArrayList<>();
// Walk down most loaded, pruning each to the max
int serversOverloaded = 0;
// flag used to fetch regions from head and tail of list, alternately
boolean fetchFromTail = false;
Map<ServerName, BalanceInfo> serverBalanceInfo = new TreeMap<>();
for (Map.Entry<ServerAndLoad, List<HRegionInfo>> server:
serversByLoad.descendingMap().entrySet()) {
ServerAndLoad sal = server.getKey();
int load = sal.getLoad();
if (load <= max) {
serverBalanceInfo.put(sal.getServerName(), new BalanceInfo(0, 0, server.getValue()));
continue;
}
serversOverloaded++;
List<HRegionInfo> regions = server.getValue();
int numToOffload = Math.min(load - max, regions.size());
// account for the out-of-band regions which were assigned to this server
// after some other region server crashed
Collections.sort(regions, riComparator);
int numTaken = 0;
for (int i = 0; i <= numToOffload; ) {
HRegionInfo hri = regions.get(i); // fetch from head
if (fetchFromTail) {
hri = regions.get(regions.size() - 1 - i);
}
i++;
// Don't rebalance special regions.
if (shouldBeOnMaster(hri)
&& masterServerName.equals(sal.getServerName())) continue;
regionsToMove.add(new RegionPlan(hri, sal.getServerName(), null));
numTaken++;
if (numTaken >= numToOffload) break;
}
serverBalanceInfo.put(sal.getServerName(),
new BalanceInfo(numToOffload, (-1)*numTaken, server.getValue()));
}
int totalNumMoved = regionsToMove.size();
// Walk down least loaded, filling each to the min
int neededRegions = 0; // number of regions needed to bring all up to min
fetchFromTail = false;
Map<ServerName, Integer> underloadedServers = new HashMap<>();
int maxToTake = numRegions - min;
for (Map.Entry<ServerAndLoad, List<HRegionInfo>> server:
serversByLoad.entrySet()) {
if (maxToTake == 0) break; // no more to take
int load = server.getKey().getLoad();
if (load >= min) {
continue; // look for other servers which haven't reached min
}
int regionsToPut = min - load;
maxToTake -= regionsToPut;
underloadedServers.put(server.getKey().getServerName(), regionsToPut);
}
// number of servers that get new regions
int serversUnderloaded = underloadedServers.size();
int incr = 1;
List<ServerName> sns =
Arrays.asList(underloadedServers.keySet().toArray(new ServerName[serversUnderloaded]));
Collections.shuffle(sns, RANDOM);
while (regionsToMove.size() > 0) {
int cnt = 0;
int i = incr > 0 ? 0 : underloadedServers.size()-1;
for (; i >= 0 && i < underloadedServers.size(); i += incr) {
if (regionsToMove.isEmpty()) break;
ServerName si = sns.get(i);
int numToTake = underloadedServers.get(si);
if (numToTake == 0) continue;
addRegionPlan(regionsToMove, fetchFromTail, si, regionsToReturn);
underloadedServers.put(si, numToTake-1);
cnt++;
BalanceInfo bi = serverBalanceInfo.get(si);
bi.setNumRegionsAdded(bi.getNumRegionsAdded()+1);
}
if (cnt == 0) break;
// iterates underloadedServers in the other direction
incr = -incr;
}
for (Integer i : underloadedServers.values()) {
// If we still want to take some, increment needed
neededRegions += i;
}
// Need to do a second pass.
// Either more regions to assign out or servers that are still underloaded
// If we need more to fill min, grab one from each most loaded until enough
if (neededRegions != 0) {
// Walk down most loaded, grabbing one from each until we get enough
for (Map.Entry<ServerAndLoad, List<HRegionInfo>> server :
serversByLoad.descendingMap().entrySet()) {
BalanceInfo balanceInfo =
serverBalanceInfo.get(server.getKey().getServerName());
int idx =
balanceInfo == null ? 0 : balanceInfo.getNextRegionForUnload();
if (idx >= server.getValue().size()) break;
HRegionInfo region = server.getValue().get(idx);
if (region.isMetaRegion()) continue; // Don't move meta regions.
regionsToMove.add(new RegionPlan(region, server.getKey().getServerName(), null));
balanceInfo.setNumRegionsAdded(balanceInfo.getNumRegionsAdded() - 1);
balanceInfo.setNextRegionForUnload(balanceInfo.getNextRegionForUnload() + 1);
totalNumMoved++;
if (--neededRegions == 0) {
// No more regions needed, done shedding
break;
}
}
}
// Now we have a set of regions that must be all assigned out
// Assign each underloaded up to the min, then if leftovers, assign to max
// Walk down least loaded, assigning to each to fill up to min
for (Map.Entry<ServerAndLoad, List<HRegionInfo>> server :
serversByLoad.entrySet()) {
int regionCount = server.getKey().getLoad();
if (regionCount >= min) break;
BalanceInfo balanceInfo = serverBalanceInfo.get(server.getKey().getServerName());
if(balanceInfo != null) {
regionCount += balanceInfo.getNumRegionsAdded();
}
if(regionCount >= min) {
continue;
}
int numToTake = min - regionCount;
int numTaken = 0;
while(numTaken < numToTake && 0 < regionsToMove.size()) {
addRegionPlan(regionsToMove, fetchFromTail,
server.getKey().getServerName(), regionsToReturn);
numTaken++;
}
}
if (min != max) {
balanceOverall(regionsToReturn, serverBalanceInfo, fetchFromTail, regionsToMove, max, min);
}
long endTime = System.currentTimeMillis();
if (!regionsToMove.isEmpty() || neededRegions != 0) {
// Emit data so can diagnose how balancer went astray.
LOG.warn("regionsToMove=" + totalNumMoved +
", numServers=" + numServers + ", serversOverloaded=" + serversOverloaded +
", serversUnderloaded=" + serversUnderloaded);
StringBuilder sb = new StringBuilder();
for (Map.Entry<ServerName, List<HRegionInfo>> e: clusterMap.entrySet()) {
if (sb.length() > 0) sb.append(", ");
sb.append(e.getKey().toString());
sb.append(" ");
sb.append(e.getValue().size());
}
LOG.warn("Input " + sb.toString());
}
// All done!
LOG.info("Done. Calculated a load balance in " + (endTime-startTime) + "ms. " +
"Moving " + totalNumMoved + " regions off of " +
serversOverloaded + " overloaded servers onto " +
serversUnderloaded + " less loaded servers");
return regionsToReturn;
}
/**
* If we need to balanceoverall, we need to add one more round to peel off one region from each max.
* Together with other regions left to be assigned, we distribute all regionToMove, to the RS
* that have less regions in whole cluster scope.
*/
public void balanceOverall(List<RegionPlan> regionsToReturn,
Map<ServerName, BalanceInfo> serverBalanceInfo, boolean fetchFromTail,
MinMaxPriorityQueue<RegionPlan> regionsToMove, int max, int min ){
// Step 1.
// A map to record the plan we have already got as status quo, in order to resolve a cyclic assignment pair,
// e.g. plan 1: A -> B, plan 2: B ->C => resolve plan1 to A -> C, remove plan2
Map<ServerName, List<Integer>> returnMap = new HashMap<>();
for (int i = 0; i < regionsToReturn.size(); i++) {
List<Integer> pos = returnMap.get(regionsToReturn.get(i).getDestination());
if (pos == null) {
pos = new ArrayList<>();
returnMap.put(regionsToReturn.get(i).getDestination(), pos);
}
pos.add(i);
}
// Step 2.
// Peel off one region from each RS which has max number of regions now.
// Each RS should have either max or min numbers of regions for this table.
for (int i = 0; i < serverLoadList.size(); i++) {
ServerAndLoad serverload = serverLoadList.get(i);
BalanceInfo balanceInfo = serverBalanceInfo.get(serverload.getServerName());
setLoad(serverLoadList, i, balanceInfo.getNumRegionsAdded());
if (balanceInfo.getHriList().size() + balanceInfo.getNumRegionsAdded() == max) {
HRegionInfo hriToPlan;
if (balanceInfo.getHriList().isEmpty()) {
LOG.debug("During balanceOverall, we found " + serverload.getServerName()
+ " has no HRegionInfo, no operation needed");
continue;
} else if (balanceInfo.getNextRegionForUnload() >= balanceInfo.getHriList().size()) {
continue;
} else {
hriToPlan = balanceInfo.getHriList().get(balanceInfo.getNextRegionForUnload());
}
RegionPlan maxPlan = new RegionPlan(hriToPlan, serverload.getServerName(), null);
regionsToMove.add(maxPlan);
setLoad(serverLoadList, i, -1);
}else if(balanceInfo.getHriList().size() + balanceInfo.getNumRegionsAdded() > max
|| balanceInfo.getHriList().size() + balanceInfo.getNumRegionsAdded() < min){
LOG.warn("Encounter incorrect region numbers after calculating move plan during balanceOverall, " +
"for this table, " + serverload.getServerName() + " originally has " + balanceInfo.getHriList().size() +
" regions and " + balanceInfo.getNumRegionsAdded() + " regions have been added. Yet, max =" +
max + ", min =" + min + ". Thus stop balance for this table"); // should not happen
return;
}
}
// Step 3. sort the ServerLoadList, the ArrayList hold overall load for each server.
// We only need to assign the regionsToMove to
// the first n = regionsToMove.size() RS that has least load.
Collections.sort(serverLoadList,new Comparator<ServerAndLoad>(){
@Override
public int compare(ServerAndLoad s1, ServerAndLoad s2) {
if(s1.getLoad() == s2.getLoad()) return 0;
else return (s1.getLoad() > s2.getLoad())? 1 : -1;
}});
// Step 4.
// Preparation before assign out all regionsToMove.
// We need to remove the plan that has the source RS equals to destination RS,
// since the source RS belongs to the least n loaded RS.
int assignLength = regionsToMove.size();
// A structure help to map ServerName to it's load and index in ServerLoadList
Map<ServerName, Pair<ServerAndLoad,Integer>> SnLoadMap = new HashMap<>();
for (int i = 0; i < serverLoadList.size(); i++) {
SnLoadMap.put(serverLoadList.get(i).getServerName(), new Pair<>(serverLoadList.get(i), i));
}
Pair<ServerAndLoad,Integer> shredLoad;
// A List to help mark the plan in regionsToMove that should be removed
List<RegionPlan> planToRemoveList = new ArrayList<>();
// A structure to record how many times a server becomes the source of a plan, from regionsToMove.
Map<ServerName, Integer> sourceMap = new HashMap<>();
// We remove one of the plan which would cause source RS equals destination RS.
// But we should keep in mind that the second plan from such RS should be kept.
for(RegionPlan plan: regionsToMove){
// the source RS's load and index in ServerLoadList
shredLoad = SnLoadMap.get(plan.getSource());
if(!sourceMap.containsKey(plan.getSource())) sourceMap.put(plan.getSource(), 0);
sourceMap.put(plan.getSource(), sourceMap.get(plan.getSource()) + 1);
if(shredLoad.getSecond() < assignLength && sourceMap.get(plan.getSource()) == 1) {
planToRemoveList.add(plan);
// While marked as to be removed, the count should be add back to the source RS
setLoad(serverLoadList, shredLoad.getSecond(), 1);
}
}
// Remove those marked plans from regionsToMove,
// we cannot direct remove them during iterating through
// regionsToMove, due to the fact that regionsToMove is a MinMaxPriorityQueue.
for(RegionPlan planToRemove : planToRemoveList){
regionsToMove.remove(planToRemove);
}
// Step 5.
// We only need to assign the regionsToMove to
// the first n = regionsToMove.size() of them, with least load.
// With this strategy adopted, we can gradually achieve the overall balance,
// while keeping table level balanced.
for(int i = 0; i < assignLength; i++){
// skip the RS that is also the source, we have removed them from regionsToMove in previous step
if(sourceMap.containsKey(serverLoadList.get(i).getServerName())) continue;
addRegionPlan(regionsToMove, fetchFromTail,
serverLoadList.get(i).getServerName(), regionsToReturn);
setLoad(serverLoadList, i, 1);
// resolve a possible cyclic assignment pair if we just produced one:
// e.g. plan1: A -> B, plan2: B -> C => resolve plan1 to A -> C and remove plan2
List<Integer> pos = returnMap.get(regionsToReturn.get(regionsToReturn.size() - 1).getSource());
if (pos != null && pos.size() != 0) {
regionsToReturn.get(pos.get(pos.size() - 1)).setDestination(
regionsToReturn.get(regionsToReturn.size() - 1).getDestination());
pos.remove(pos.size() - 1);
regionsToReturn.remove(regionsToReturn.size() - 1);
}
}
// Done balance overall
}
/**
* Add a region from the head or tail to the List of regions to return.
*/
private void addRegionPlan(final MinMaxPriorityQueue<RegionPlan> regionsToMove,
final boolean fetchFromTail, final ServerName sn, List<RegionPlan> regionsToReturn) {
RegionPlan rp = null;
if (!fetchFromTail) rp = regionsToMove.remove();
else rp = regionsToMove.removeLast();
rp.setDestination(sn);
regionsToReturn.add(rp);
}
@Override
public List<RegionPlan> balanceCluster(TableName tableName,
Map<ServerName, List<HRegionInfo>> clusterState) throws HBaseIOException {
LOG.debug("Start Generate Balance plan for table: " + tableName);
return balanceCluster(clusterState);
}
}