Examples with ServerName org.apache.hadoop.hbase.ServerName used on opensource projects

Example 61 with ServerName

use of org.apache.hadoop.hbase.ServerName in project hbase by apache.

the class FavoredNodesManager method updateFavoredNodes.

public synchronized void updateFavoredNodes(Map<HRegionInfo, List<ServerName>> regionFNMap) throws IOException {
    Map<HRegionInfo, List<ServerName>> regionToFavoredNodes = new HashMap<>();
    for (Map.Entry<HRegionInfo, List<ServerName>> entry : regionFNMap.entrySet()) {
        HRegionInfo regionInfo = entry.getKey();
        List<ServerName> servers = entry.getValue();
        /*
       * None of the following error conditions should happen. If it does, there is an issue with
       * favored nodes generation or the regions its called on.
       */
        if (servers.size() != Sets.newHashSet(servers).size()) {
            throw new IOException("Duplicates found: " + servers);
        }
        if (!isFavoredNodeApplicable(regionInfo)) {
            throw new IOException("Can't update FN for a un-applicable region: " + regionInfo.getRegionNameAsString() + " with " + servers);
        }
        if (servers.size() != FavoredNodeAssignmentHelper.FAVORED_NODES_NUM) {
            throw new IOException("At least " + FavoredNodeAssignmentHelper.FAVORED_NODES_NUM + " favored nodes should be present for region : " + regionInfo.getEncodedName() + " current FN servers:" + servers);
        }
        List<ServerName> serversWithNoStartCodes = Lists.newArrayList();
        for (ServerName sn : servers) {
            if (sn.getStartcode() == ServerName.NON_STARTCODE) {
                serversWithNoStartCodes.add(sn);
            } else {
                serversWithNoStartCodes.add(ServerName.valueOf(sn.getHostname(), sn.getPort(), ServerName.NON_STARTCODE));
            }
        }
        regionToFavoredNodes.put(regionInfo, serversWithNoStartCodes);
    }
    // Lets do a bulk update to meta since that reduces the RPC's
    FavoredNodeAssignmentHelper.updateMetaWithFavoredNodesInfo(regionToFavoredNodes, masterServices.getConnection());
    deleteFavoredNodesForRegions(regionToFavoredNodes.keySet());
    for (Map.Entry<HRegionInfo, List<ServerName>> entry : regionToFavoredNodes.entrySet()) {
        HRegionInfo regionInfo = entry.getKey();
        List<ServerName> serversWithNoStartCodes = entry.getValue();
        globalFavoredNodesAssignmentPlan.updateFavoredNodesMap(regionInfo, serversWithNoStartCodes);
        addToReplicaLoad(regionInfo, serversWithNoStartCodes);
    }
}

Example 62 with ServerName

use of org.apache.hadoop.hbase.ServerName in project hbase by apache.

the class TakeSnapshotHandler method process.

/**
   * Execute the core common portions of taking a snapshot. The {@link #snapshotRegions(List)}
   * call should get implemented for each snapshot flavor.
   */
@Override
@edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "REC_CATCH_EXCEPTION", justification = "Intentional")
public void process() {
    String msg = "Running " + snapshot.getType() + " table snapshot " + snapshot.getName() + " " + eventType + " on table " + snapshotTable;
    LOG.info(msg);
    ReentrantLock lock = snapshotManager.getLocks().acquireLock(snapshot.getName());
    status.setStatus(msg);
    try {
        // If regions move after this meta scan, the region specific snapshot should fail, triggering
        // an external exception that gets captured here.
        // write down the snapshot info in the working directory
        SnapshotDescriptionUtils.writeSnapshotInfo(snapshot, workingDir, fs);
        snapshotManifest.addTableDescriptor(this.htd);
        monitor.rethrowException();
        List<Pair<HRegionInfo, ServerName>> regionsAndLocations;
        if (TableName.META_TABLE_NAME.equals(snapshotTable)) {
            regionsAndLocations = new MetaTableLocator().getMetaRegionsAndLocations(server.getZooKeeper());
        } else {
            regionsAndLocations = MetaTableAccessor.getTableRegionsAndLocations(server.getConnection(), snapshotTable, false);
        }
        // run the snapshot
        snapshotRegions(regionsAndLocations);
        monitor.rethrowException();
        // extract each pair to separate lists
        Set<String> serverNames = new HashSet<>();
        for (Pair<HRegionInfo, ServerName> p : regionsAndLocations) {
            if (p != null && p.getFirst() != null && p.getSecond() != null) {
                HRegionInfo hri = p.getFirst();
                if (hri.isOffline() && (hri.isSplit() || hri.isSplitParent()))
                    continue;
                serverNames.add(p.getSecond().toString());
            }
        }
        // flush the in-memory state, and write the single manifest
        status.setStatus("Consolidate snapshot: " + snapshot.getName());
        snapshotManifest.consolidate();
        // verify the snapshot is valid
        status.setStatus("Verifying snapshot: " + snapshot.getName());
        verifier.verifySnapshot(this.workingDir, serverNames);
        // complete the snapshot, atomically moving from tmp to .snapshot dir.
        completeSnapshot(this.snapshotDir, this.workingDir, this.fs);
        msg = "Snapshot " + snapshot.getName() + " of table " + snapshotTable + " completed";
        status.markComplete(msg);
        LOG.info(msg);
        metricsSnapshot.addSnapshot(status.getCompletionTimestamp() - status.getStartTime());
    } catch (Exception e) {
        // FindBugs: REC_CATCH_EXCEPTION
        status.abort("Failed to complete snapshot " + snapshot.getName() + " on table " + snapshotTable + " because " + e.getMessage());
        String reason = "Failed taking snapshot " + ClientSnapshotDescriptionUtils.toString(snapshot) + " due to exception:" + e.getMessage();
        LOG.error(reason, e);
        ForeignException ee = new ForeignException(reason, e);
        monitor.receive(ee);
        // need to mark this completed to close off and allow cleanup to happen.
        cancel(reason);
    } finally {
        LOG.debug("Launching cleanup of working dir:" + workingDir);
        try {
            // it.
            if (fs.exists(workingDir) && !this.fs.delete(workingDir, true)) {
                LOG.error("Couldn't delete snapshot working directory:" + workingDir);
            }
        } catch (IOException e) {
            LOG.error("Couldn't delete snapshot working directory:" + workingDir);
        }
        lock.unlock();
        tableLock.release();
    }
}

Example 63 with ServerName

use of org.apache.hadoop.hbase.ServerName in project hbase by apache.

the class DisabledTableSnapshotHandler method snapshotRegions.

// TODO consider parallelizing these operations since they are independent. Right now its just
// easier to keep them serial though
@Override
public void snapshotRegions(List<Pair<HRegionInfo, ServerName>> regionsAndLocations) throws IOException, KeeperException {
    try {
        // 1. get all the regions hosting this table.
        // extract each pair to separate lists
        Set<HRegionInfo> regions = new HashSet<>();
        for (Pair<HRegionInfo, ServerName> p : regionsAndLocations) {
            // Don't include non-default regions
            HRegionInfo hri = p.getFirst();
            if (RegionReplicaUtil.isDefaultReplica(hri)) {
                regions.add(hri);
            }
        }
        // handle the mob files if any.
        boolean mobEnabled = MobUtils.hasMobColumns(htd);
        if (mobEnabled) {
            // snapshot the mob files as a offline region.
            HRegionInfo mobRegionInfo = MobUtils.getMobRegionInfo(htd.getTableName());
            regions.add(mobRegionInfo);
        }
        // 2. for each region, write all the info to disk
        String msg = "Starting to write region info and WALs for regions for offline snapshot:" + ClientSnapshotDescriptionUtils.toString(snapshot);
        LOG.info(msg);
        status.setStatus(msg);
        ThreadPoolExecutor exec = SnapshotManifest.createExecutor(conf, "DisabledTableSnapshot");
        try {
            ModifyRegionUtils.editRegions(exec, regions, new ModifyRegionUtils.RegionEditTask() {

                @Override
                public void editRegion(final HRegionInfo regionInfo) throws IOException {
                    snapshotManifest.addRegion(FSUtils.getTableDir(rootDir, snapshotTable), regionInfo);
                }
            });
        } finally {
            exec.shutdown();
        }
    } catch (Exception e) {
        // make sure we capture the exception to propagate back to the client later
        String reason = "Failed snapshot " + ClientSnapshotDescriptionUtils.toString(snapshot) + " due to exception:" + e.getMessage();
        ForeignException ee = new ForeignException(reason, e);
        monitor.receive(ee);
        status.abort("Snapshot of table: " + snapshotTable + " failed because " + e.getMessage());
    } finally {
        LOG.debug("Marking snapshot" + ClientSnapshotDescriptionUtils.toString(snapshot) + " as finished.");
    }
}

Example 64 with ServerName

use of org.apache.hadoop.hbase.ServerName in project hbase by apache.

the class ServerCrashProcedure method calcRegionsToAssign.

/**
   * Figure out what we need to assign. Should be idempotent.
   * @param env
   * @return List of calculated regions to assign; may be empty or null.
   * @throws IOException
   */
private List<HRegionInfo> calcRegionsToAssign(final MasterProcedureEnv env) throws IOException {
    AssignmentManager am = env.getMasterServices().getAssignmentManager();
    List<HRegionInfo> regionsToAssignAggregator = new ArrayList<>();
    int replicaCount = env.getMasterConfiguration().getInt(HConstants.META_REPLICAS_NUM, HConstants.DEFAULT_META_REPLICA_NUM);
    for (int i = 1; i < replicaCount; i++) {
        HRegionInfo metaHri = RegionReplicaUtil.getRegionInfoForReplica(HRegionInfo.FIRST_META_REGIONINFO, i);
        if (am.isCarryingMetaReplica(this.serverName, metaHri)) {
            if (LOG.isDebugEnabled()) {
                LOG.debug("Reassigning meta replica" + metaHri + " that was on " + this.serverName);
            }
            regionsToAssignAggregator.add(metaHri);
        }
    }
    // Clean out anything in regions in transition.
    List<HRegionInfo> regionsInTransition = am.cleanOutCrashedServerReferences(serverName);
    if (LOG.isDebugEnabled()) {
        LOG.debug("Reassigning " + size(this.regionsOnCrashedServer) + " region(s) that " + (serverName == null ? "null" : serverName) + " was carrying (and " + regionsInTransition.size() + " regions(s) that were opening on this server)");
    }
    regionsToAssignAggregator.addAll(regionsInTransition);
    // Iterate regions that were on this server and figure which of these we need to reassign
    if (this.regionsOnCrashedServer != null && !this.regionsOnCrashedServer.isEmpty()) {
        RegionStates regionStates = am.getRegionStates();
        for (HRegionInfo hri : this.regionsOnCrashedServer) {
            if (regionsInTransition.contains(hri))
                continue;
            String encodedName = hri.getEncodedName();
            Lock lock = am.acquireRegionLock(encodedName);
            try {
                RegionState rit = regionStates.getRegionTransitionState(hri);
                if (processDeadRegion(hri, am)) {
                    ServerName addressFromAM = regionStates.getRegionServerOfRegion(hri);
                    if (addressFromAM != null && !addressFromAM.equals(this.serverName)) {
                        // If this region is in transition on the dead server, it must be
                        // opening or pending_open, which should have been covered by
                        // AM#cleanOutCrashedServerReferences
                        LOG.info("Skip assigning " + hri.getRegionNameAsString() + " because opened on " + addressFromAM.getServerName());
                        continue;
                    }
                    if (rit != null) {
                        if (rit.getServerName() != null && !rit.isOnServer(this.serverName)) {
                            // Skip regions that are in transition on other server
                            LOG.info("Skip assigning region in transition on other server" + rit);
                            continue;
                        }
                        LOG.info("Reassigning region " + rit + " and clearing zknode if exists");
                        regionStates.updateRegionState(hri, RegionState.State.OFFLINE);
                    } else if (regionStates.isRegionInState(hri, RegionState.State.SPLITTING_NEW, RegionState.State.MERGING_NEW)) {
                        regionStates.updateRegionState(hri, RegionState.State.OFFLINE);
                    }
                    regionsToAssignAggregator.add(hri);
                // TODO: The below else if is different in branch-1 from master branch.
                } else if (rit != null) {
                    if ((rit.isClosing() || rit.isFailedClose() || rit.isOffline()) && am.getTableStateManager().isTableState(hri.getTable(), TableState.State.DISABLED, TableState.State.DISABLING) || am.getReplicasToClose().contains(hri)) {
                        // If the table was partially disabled and the RS went down, we should clear the
                        // RIT and remove the node for the region.
                        // The rit that we use may be stale in case the table was in DISABLING state
                        // but though we did assign we will not be clearing the znode in CLOSING state.
                        // Doing this will have no harm. See HBASE-5927
                        regionStates.updateRegionState(hri, RegionState.State.OFFLINE);
                        am.offlineDisabledRegion(hri);
                    } else {
                        LOG.warn("THIS SHOULD NOT HAPPEN: unexpected region in transition " + rit + " not to be assigned by SSH of server " + serverName);
                    }
                }
            } finally {
                lock.unlock();
            }
        }
    }
    return regionsToAssignAggregator;
}

Example 65 with ServerName

use of org.apache.hadoop.hbase.ServerName in project hbase by apache.

the class Canary method sniff.

/*
   * Loops over regions that owns this table, and output some information abouts the state.
   */
private static List<Future<Void>> sniff(final Admin admin, final Sink sink, HTableDescriptor tableDesc, ExecutorService executor, TaskType taskType, boolean rawScanEnabled) throws Exception {
    if (LOG.isDebugEnabled()) {
        LOG.debug(String.format("reading list of regions for table %s", tableDesc.getTableName()));
    }
    Table table = null;
    try {
        table = admin.getConnection().getTable(tableDesc.getTableName());
    } catch (TableNotFoundException e) {
        return new ArrayList<>();
    } finally {
        if (table != null) {
            table.close();
        }
    }
    List<RegionTask> tasks = new ArrayList<>();
    RegionLocator regionLocator = null;
    try {
        regionLocator = admin.getConnection().getRegionLocator(tableDesc.getTableName());
        for (HRegionLocation location : regionLocator.getAllRegionLocations()) {
            ServerName rs = location.getServerName();
            HRegionInfo region = location.getRegionInfo();
            tasks.add(new RegionTask(admin.getConnection(), region, rs, sink, taskType, rawScanEnabled));
        }
    } finally {
        if (regionLocator != null) {
            regionLocator.close();
        }
    }
    return executor.invokeAll(tasks);
}