Examples with AssignmentManager org.apache.hadoop.hbase.master.assignment.AssignmentManager used on opensource projects

Example 1 with AssignmentManager

use of org.apache.hadoop.hbase.master.assignment.AssignmentManager in project hbase by apache.

the class HMaster method getClusterMetricsWithoutCoprocessor.

public ClusterMetrics getClusterMetricsWithoutCoprocessor(EnumSet<Option> options) throws InterruptedIOException {
    ClusterMetricsBuilder builder = ClusterMetricsBuilder.newBuilder();
    // we return all information to client if the list of Option is empty.
    if (options.isEmpty()) {
        options = EnumSet.allOf(Option.class);
    }
    // TASKS and/or LIVE_SERVERS will populate this map, which will be given to the builder if
    // not null after option processing completes.
    Map<ServerName, ServerMetrics> serverMetricsMap = null;
    for (Option opt : options) {
        switch(opt) {
            case HBASE_VERSION:
                builder.setHBaseVersion(VersionInfo.getVersion());
                break;
            case CLUSTER_ID:
                builder.setClusterId(getClusterId());
                break;
            case MASTER:
                builder.setMasterName(getServerName());
                break;
            case BACKUP_MASTERS:
                builder.setBackerMasterNames(getBackupMasters());
                break;
            case TASKS:
                {
                    // Master tasks
                    builder.setMasterTasks(TaskMonitor.get().getTasks().stream().map(task -> ServerTaskBuilder.newBuilder().setDescription(task.getDescription()).setStatus(task.getStatus()).setState(ServerTask.State.valueOf(task.getState().name())).setStartTime(task.getStartTime()).setCompletionTime(task.getCompletionTimestamp()).build()).collect(Collectors.toList()));
                    // Add entries to serverMetricsMap for all live servers, if we haven't already done so
                    if (serverMetricsMap == null) {
                        serverMetricsMap = getOnlineServers();
                    }
                    break;
                }
            case LIVE_SERVERS:
                {
                    // Add entries to serverMetricsMap for all live servers, if we haven't already done so
                    if (serverMetricsMap == null) {
                        serverMetricsMap = getOnlineServers();
                    }
                    break;
                }
            case DEAD_SERVERS:
                {
                    if (serverManager != null) {
                        builder.setDeadServerNames(new ArrayList<>(serverManager.getDeadServers().copyServerNames()));
                    }
                    break;
                }
            case MASTER_COPROCESSORS:
                {
                    if (cpHost != null) {
                        builder.setMasterCoprocessorNames(Arrays.asList(getMasterCoprocessors()));
                    }
                    break;
                }
            case REGIONS_IN_TRANSITION:
                {
                    if (assignmentManager != null) {
                        builder.setRegionsInTransition(assignmentManager.getRegionStates().getRegionsStateInTransition());
                    }
                    break;
                }
            case BALANCER_ON:
                {
                    if (loadBalancerTracker != null) {
                        builder.setBalancerOn(loadBalancerTracker.isBalancerOn());
                    }
                    break;
                }
            case MASTER_INFO_PORT:
                {
                    if (infoServer != null) {
                        builder.setMasterInfoPort(infoServer.getPort());
                    }
                    break;
                }
            case SERVERS_NAME:
                {
                    if (serverManager != null) {
                        builder.setServerNames(serverManager.getOnlineServersList());
                    }
                    break;
                }
            case TABLE_TO_REGIONS_COUNT:
                {
                    if (isActiveMaster() && isInitialized() && assignmentManager != null) {
                        try {
                            Map<TableName, RegionStatesCount> tableRegionStatesCountMap = new HashMap<>();
                            Map<String, TableDescriptor> tableDescriptorMap = getTableDescriptors().getAll();
                            for (TableDescriptor tableDescriptor : tableDescriptorMap.values()) {
                                TableName tableName = tableDescriptor.getTableName();
                                RegionStatesCount regionStatesCount = assignmentManager.getRegionStatesCount(tableName);
                                tableRegionStatesCountMap.put(tableName, regionStatesCount);
                            }
                            builder.setTableRegionStatesCount(tableRegionStatesCountMap);
                        } catch (IOException e) {
                            LOG.error("Error while populating TABLE_TO_REGIONS_COUNT for Cluster Metrics..", e);
                        }
                    }
                    break;
                }
        }
    }
    if (serverMetricsMap != null) {
        builder.setLiveServerMetrics(serverMetricsMap);
    }
    return builder.build();
}

Example 2 with AssignmentManager

use of org.apache.hadoop.hbase.master.assignment.AssignmentManager in project hbase by apache.

the class CloneSnapshotProcedure method executeFromState.

@Override
protected Flow executeFromState(final MasterProcedureEnv env, final CloneSnapshotState state) throws InterruptedException {
    LOG.trace("{} execute state={}", this, state);
    try {
        switch(state) {
            case CLONE_SNAPSHOT_PRE_OPERATION:
                // Verify if we can clone the table
                prepareClone(env);
                preCloneSnapshot(env);
                setNextState(CloneSnapshotState.CLONE_SNAPSHOT_WRITE_FS_LAYOUT);
                break;
            case CLONE_SNAPSHOT_WRITE_FS_LAYOUT:
                updateTableDescriptorWithSFT();
                newRegions = createFilesystemLayout(env, tableDescriptor, newRegions);
                env.getMasterServices().getTableDescriptors().update(tableDescriptor, true);
                setNextState(CloneSnapshotState.CLONE_SNAPSHOT_ADD_TO_META);
                break;
            case CLONE_SNAPSHOT_ADD_TO_META:
                addRegionsToMeta(env);
                setNextState(CloneSnapshotState.CLONE_SNAPSHOT_ASSIGN_REGIONS);
                break;
            case CLONE_SNAPSHOT_ASSIGN_REGIONS:
                CreateTableProcedure.setEnablingState(env, getTableName());
                // Separate newRegions to split regions and regions to assign
                List<RegionInfo> splitRegions = new ArrayList<>();
                List<RegionInfo> regionsToAssign = new ArrayList<>();
                newRegions.forEach(ri -> {
                    if (ri.isOffline() && (ri.isSplit() || ri.isSplitParent())) {
                        splitRegions.add(ri);
                    } else {
                        regionsToAssign.add(ri);
                    }
                });
                // For split regions, add them to RegionStates
                AssignmentManager am = env.getAssignmentManager();
                splitRegions.forEach(ri -> am.getRegionStates().updateRegionState(ri, RegionState.State.SPLIT));
                addChildProcedure(env.getAssignmentManager().createRoundRobinAssignProcedures(regionsToAssign));
                setNextState(CloneSnapshotState.CLONE_SNAPSHOT_UPDATE_DESC_CACHE);
                break;
            case CLONE_SNAPSHOT_UPDATE_DESC_CACHE:
                // XXX: this stage should be named as set table enabled, as now we will cache the
                // descriptor after writing fs layout.
                CreateTableProcedure.setEnabledState(env, getTableName());
                setNextState(CloneSnapshotState.CLONE_SNAPHOST_RESTORE_ACL);
                break;
            case CLONE_SNAPHOST_RESTORE_ACL:
                restoreSnapshotAcl(env);
                setNextState(CloneSnapshotState.CLONE_SNAPSHOT_POST_OPERATION);
                break;
            case CLONE_SNAPSHOT_POST_OPERATION:
                postCloneSnapshot(env);
                MetricsSnapshot metricsSnapshot = new MetricsSnapshot();
                metricsSnapshot.addSnapshotClone(getMonitorStatus().getCompletionTimestamp() - getMonitorStatus().getStartTime());
                getMonitorStatus().markComplete("Clone snapshot '" + snapshot.getName() + "' completed!");
                return Flow.NO_MORE_STATE;
            default:
                throw new UnsupportedOperationException("unhandled state=" + state);
        }
    } catch (IOException e) {
        if (isRollbackSupported(state)) {
            setFailure("master-clone-snapshot", e);
        } else {
            LOG.warn("Retriable error trying to clone snapshot=" + snapshot.getName() + " to table=" + getTableName() + " state=" + state, e);
        }
    }
    return Flow.HAS_MORE_STATE;
}

Example 3 with AssignmentManager

use of org.apache.hadoop.hbase.master.assignment.AssignmentManager in project hbase by apache.

the class MasterDumpServlet method dumpRIT.

private void dumpRIT(HMaster master, PrintWriter out) {
    AssignmentManager am = master.getAssignmentManager();
    if (am == null) {
        out.println("AssignmentManager is not initialized");
        return;
    }
    for (RegionStateNode rs : am.getRegionsInTransition()) {
        String rid = rs.getRegionInfo().getEncodedName();
        out.println("Region " + rid + ": " + rs.toDescriptiveString());
    }
}

Example 4 with AssignmentManager

use of org.apache.hadoop.hbase.master.assignment.AssignmentManager in project hbase by apache.

the class ServerCrashProcedure method assignRegions.

/**
 * Assign the regions on the crashed RS to other Rses.
 * <p/>
 * In this method we will go through all the RegionStateNodes of the give regions to find out
 * whether there is already an TRSP for the region, if so we interrupt it and let it retry on
 * other server, otherwise we will schedule a TRSP to bring the region online.
 * <p/>
 * We will also check whether the table for a region is enabled, if not, we will skip assigning
 * it.
 */
private void assignRegions(MasterProcedureEnv env, List<RegionInfo> regions) throws IOException {
    AssignmentManager am = env.getMasterServices().getAssignmentManager();
    boolean retainAssignment = env.getMasterConfiguration().getBoolean(MASTER_SCP_RETAIN_ASSIGNMENT, DEFAULT_MASTER_SCP_RETAIN_ASSIGNMENT);
    for (RegionInfo region : regions) {
        RegionStateNode regionNode = am.getRegionStates().getOrCreateRegionStateNode(region);
        regionNode.lock();
        try {
            // in the way of our clearing out 'Unknown Servers'.
            if (!isMatchingRegionLocation(regionNode)) {
                // double checking here to confirm that we do not skip assignment incorrectly.
                if (!am.isRunning()) {
                    throw new DoNotRetryIOException("AssignmentManager has been stopped, can not process assignment any more");
                }
                LOG.info("{} found {} whose regionLocation no longer matches {}, skipping assign...", this, regionNode, serverName);
                continue;
            }
            if (regionNode.getProcedure() != null) {
                LOG.info("{} found RIT {}; {}", this, regionNode.getProcedure(), regionNode);
                regionNode.getProcedure().serverCrashed(env, regionNode, getServerName(), !retainAssignment);
                continue;
            }
            if (env.getMasterServices().getTableStateManager().isTableState(regionNode.getTable(), TableState.State.DISABLING)) {
                // We need to change the state here otherwise the TRSP scheduled by DTP will try to
                // close the region from a dead server and will never succeed. Please see HBASE-23636
                // for more details.
                env.getAssignmentManager().regionClosedAbnormally(regionNode);
                LOG.info("{} found table disabling for region {}, set it state to ABNORMALLY_CLOSED.", this, regionNode);
                continue;
            }
            if (env.getMasterServices().getTableStateManager().isTableState(regionNode.getTable(), TableState.State.DISABLED)) {
                // This should not happen, table disabled but has regions on server.
                LOG.warn("Found table disabled for region {}, procDetails: {}", regionNode, this);
                continue;
            }
            TransitRegionStateProcedure proc = TransitRegionStateProcedure.assign(env, region, !retainAssignment, null);
            regionNode.setProcedure(proc);
            addChildProcedure(proc);
        } finally {
            regionNode.unlock();
        }
    }
}

Example 5 with AssignmentManager

use of org.apache.hadoop.hbase.master.assignment.AssignmentManager in project hbase by apache.

the class ServerCrashProcedure method zkCoordinatedSplitLogs.

/**
 * Split logs using 'classic' zk-based coordination.
 * Superceded by procedure-based WAL splitting.
 * @see #createSplittingWalProcedures(MasterProcedureEnv, boolean)
 */
private void zkCoordinatedSplitLogs(final MasterProcedureEnv env) throws IOException {
    LOG.debug("Splitting WALs {}", this);
    MasterWalManager mwm = env.getMasterServices().getMasterWalManager();
    AssignmentManager am = env.getMasterServices().getAssignmentManager();
    // TODO: For Matteo. Below BLOCKs!!!! Redo so can relinquish executor while it is running.
    // PROBLEM!!! WE BLOCK HERE. Can block for hours if hundreds of WALs to split and hundreds
    // of SCPs running because big cluster crashed down.
    am.getRegionStates().logSplitting(this.serverName);
    mwm.splitLog(this.serverName);
    if (!carryingMeta) {
        mwm.archiveMetaLog(this.serverName);
    }
    am.getRegionStates().logSplit(this.serverName);
    LOG.debug("Done splitting WALs {}", this);
}