Examples with TabletLocation org.apache.accumulo.core.client.impl.TabletLocator.TabletLocation used on opensource projects

Example 1 with TabletLocation

use of org.apache.accumulo.core.client.impl.TabletLocator.TabletLocation in project accumulo by apache.

the class BulkImporter method findOverlappingTablets.

public static List<TabletLocation> findOverlappingTablets(ClientContext context, VolumeManager vm, TabletLocator locator, Path file, Text startRow, Text endRow) throws Exception {
    List<TabletLocation> result = new ArrayList<>();
    Collection<ByteSequence> columnFamilies = Collections.emptyList();
    String filename = file.toString();
    // log.debug(filename + " finding overlapping tablets " + startRow + " -> " + endRow);
    FileSystem fs = vm.getVolumeByPath(file).getFileSystem();
    try (FileSKVIterator reader = FileOperations.getInstance().newReaderBuilder().forFile(filename, fs, fs.getConf()).withTableConfiguration(context.getConfiguration()).seekToBeginning().build()) {
        Text row = startRow;
        if (row == null)
            row = new Text();
        while (true) {
            // log.debug(filename + " Seeking to row " + row);
            reader.seek(new Range(row, null), columnFamilies, false);
            if (!reader.hasTop()) {
                // log.debug(filename + " not found");
                break;
            }
            row = reader.getTopKey().getRow();
            TabletLocation tabletLocation = locator.locateTablet(context, row, false, true);
            // log.debug(filename + " found row " + row + " at location " + tabletLocation);
            result.add(tabletLocation);
            row = tabletLocation.tablet_extent.getEndRow();
            if (row != null && (endRow == null || row.compareTo(endRow) < 0)) {
                row = new Text(row);
                row.append(byte0, 0, byte0.length);
            } else
                break;
        }
    }
    // log.debug(filename + " to be sent to " + result);
    return result;
}

Example 2 with TabletLocation

use of org.apache.accumulo.core.client.impl.TabletLocator.TabletLocation in project accumulo by apache.

the class BulkImporter method estimateSizes.

private Map<Path, List<AssignmentInfo>> estimateSizes(final AccumuloConfiguration acuConf, final Configuration conf, final VolumeManager vm, Map<Path, List<TabletLocation>> assignments, Collection<Path> paths, int numThreads) {
    long t1 = System.currentTimeMillis();
    final Map<Path, Long> mapFileSizes = new TreeMap<>();
    try {
        for (Path path : paths) {
            FileSystem fs = vm.getVolumeByPath(path).getFileSystem();
            mapFileSizes.put(path, fs.getContentSummary(path).getLength());
        }
    } catch (IOException e) {
        log.error("Failed to get map files in for {}: {}", paths, e.getMessage(), e);
        throw new RuntimeException(e);
    }
    final Map<Path, List<AssignmentInfo>> ais = Collections.synchronizedMap(new TreeMap<Path, List<AssignmentInfo>>());
    ExecutorService threadPool = Executors.newFixedThreadPool(numThreads, new NamingThreadFactory("estimateSizes"));
    for (final Entry<Path, List<TabletLocation>> entry : assignments.entrySet()) {
        if (entry.getValue().size() == 1) {
            TabletLocation tabletLocation = entry.getValue().get(0);
            // if the tablet completely contains the map file, there is no
            // need to estimate its
            // size
            ais.put(entry.getKey(), Collections.singletonList(new AssignmentInfo(tabletLocation.tablet_extent, mapFileSizes.get(entry.getKey()))));
            continue;
        }
        Runnable estimationTask = new Runnable() {

            @Override
            public void run() {
                Map<KeyExtent, Long> estimatedSizes = null;
                try {
                    estimatedSizes = FileUtil.estimateSizes(acuConf, entry.getKey(), mapFileSizes.get(entry.getKey()), extentsOf(entry.getValue()), conf, vm);
                } catch (IOException e) {
                    log.warn("Failed to estimate map file sizes {}", e.getMessage());
                }
                if (estimatedSizes == null) {
                    // estimation failed, do a simple estimation
                    estimatedSizes = new TreeMap<>();
                    long estSize = (long) (mapFileSizes.get(entry.getKey()) / (double) entry.getValue().size());
                    for (TabletLocation tl : entry.getValue()) estimatedSizes.put(tl.tablet_extent, estSize);
                }
                List<AssignmentInfo> assignmentInfoList = new ArrayList<>(estimatedSizes.size());
                for (Entry<KeyExtent, Long> entry2 : estimatedSizes.entrySet()) assignmentInfoList.add(new AssignmentInfo(entry2.getKey(), entry2.getValue()));
                ais.put(entry.getKey(), assignmentInfoList);
            }
        };
        threadPool.submit(new TraceRunnable(new LoggingRunnable(log, estimationTask)));
    }
    threadPool.shutdown();
    while (!threadPool.isTerminated()) {
        try {
            threadPool.awaitTermination(60, TimeUnit.SECONDS);
        } catch (InterruptedException e) {
            log.error("Encountered InterruptedException while waiting for the threadPool to terminate.", e);
            throw new RuntimeException(e);
        }
    }
    long t2 = System.currentTimeMillis();
    log.debug(String.format("Estimated map files sizes in %6.2f secs", (t2 - t1) / 1000.0));
    return ais;
}

Example 3 with TabletLocation

use of org.apache.accumulo.core.client.impl.TabletLocator.TabletLocation in project accumulo by apache.

the class BulkImporter method importFiles.

public AssignmentStats importFiles(List<String> files, Path failureDir) throws IOException, AccumuloException, AccumuloSecurityException, ThriftTableOperationException {
    int numThreads = context.getConfiguration().getCount(Property.TSERV_BULK_PROCESS_THREADS);
    int numAssignThreads = context.getConfiguration().getCount(Property.TSERV_BULK_ASSIGNMENT_THREADS);
    timer = new StopWatch<>(Timers.class);
    timer.start(Timers.TOTAL);
    Configuration conf = CachedConfiguration.getInstance();
    VolumeManagerImpl.get(context.getConfiguration());
    final VolumeManager fs = VolumeManagerImpl.get(context.getConfiguration());
    Set<Path> paths = new HashSet<>();
    for (String file : files) {
        paths.add(new Path(file));
    }
    AssignmentStats assignmentStats = new AssignmentStats(paths.size());
    final Map<Path, List<KeyExtent>> completeFailures = Collections.synchronizedSortedMap(new TreeMap<Path, List<KeyExtent>>());
    ClientService.Client client = null;
    final TabletLocator locator = TabletLocator.getLocator(context, Table.ID.of(tableId));
    try {
        final Map<Path, List<TabletLocation>> assignments = Collections.synchronizedSortedMap(new TreeMap<Path, List<TabletLocation>>());
        timer.start(Timers.EXAMINE_MAP_FILES);
        ExecutorService threadPool = Executors.newFixedThreadPool(numThreads, new NamingThreadFactory("findOverlapping"));
        for (Path path : paths) {
            final Path mapFile = path;
            Runnable getAssignments = new Runnable() {

                @Override
                public void run() {
                    List<TabletLocation> tabletsToAssignMapFileTo = Collections.emptyList();
                    try {
                        tabletsToAssignMapFileTo = findOverlappingTablets(context, fs, locator, mapFile);
                    } catch (Exception ex) {
                        log.warn("Unable to find tablets that overlap file " + mapFile.toString(), ex);
                    }
                    log.debug("Map file {} found to overlap {} tablets", mapFile, tabletsToAssignMapFileTo.size());
                    if (tabletsToAssignMapFileTo.size() == 0) {
                        List<KeyExtent> empty = Collections.emptyList();
                        completeFailures.put(mapFile, empty);
                    } else
                        assignments.put(mapFile, tabletsToAssignMapFileTo);
                }
            };
            threadPool.submit(new TraceRunnable(new LoggingRunnable(log, getAssignments)));
        }
        threadPool.shutdown();
        while (!threadPool.isTerminated()) {
            try {
                threadPool.awaitTermination(60, TimeUnit.SECONDS);
            } catch (InterruptedException e) {
                throw new RuntimeException(e);
            }
        }
        timer.stop(Timers.EXAMINE_MAP_FILES);
        assignmentStats.attemptingAssignments(assignments);
        Map<Path, List<KeyExtent>> assignmentFailures = assignMapFiles(context, conf, fs, tableId, assignments, paths, numAssignThreads, numThreads);
        assignmentStats.assignmentsFailed(assignmentFailures);
        Map<Path, Integer> failureCount = new TreeMap<>();
        for (Entry<Path, List<KeyExtent>> entry : assignmentFailures.entrySet()) failureCount.put(entry.getKey(), 1);
        long sleepTime = 2 * 1000;
        while (assignmentFailures.size() > 0) {
            sleepTime = Math.min(sleepTime * 2, 60 * 1000);
            locator.invalidateCache();
            // assumption about assignment failures is that it caused by a split
            // happening or a missing location
            // 
            // for splits we need to find children key extents that cover the
            // same key range and are contiguous (no holes, no overlap)
            timer.start(Timers.SLEEP);
            sleepUninterruptibly(sleepTime, TimeUnit.MILLISECONDS);
            timer.stop(Timers.SLEEP);
            log.debug("Trying to assign {} map files that previously failed on some key extents", assignmentFailures.size());
            assignments.clear();
            // assign to
            for (Entry<Path, List<KeyExtent>> entry : assignmentFailures.entrySet()) {
                Iterator<KeyExtent> keListIter = entry.getValue().iterator();
                List<TabletLocation> tabletsToAssignMapFileTo = new ArrayList<>();
                while (keListIter.hasNext()) {
                    KeyExtent ke = keListIter.next();
                    timer.start(Timers.QUERY_METADATA);
                    try {
                        tabletsToAssignMapFileTo.addAll(findOverlappingTablets(context, fs, locator, entry.getKey(), ke));
                        keListIter.remove();
                    } catch (Exception ex) {
                        log.warn("Exception finding overlapping tablets, will retry tablet " + ke, ex);
                    }
                    timer.stop(Timers.QUERY_METADATA);
                }
                if (tabletsToAssignMapFileTo.size() > 0)
                    assignments.put(entry.getKey(), tabletsToAssignMapFileTo);
            }
            assignmentStats.attemptingAssignments(assignments);
            Map<Path, List<KeyExtent>> assignmentFailures2 = assignMapFiles(context, conf, fs, tableId, assignments, paths, numAssignThreads, numThreads);
            assignmentStats.assignmentsFailed(assignmentFailures2);
            // merge assignmentFailures2 into assignmentFailures
            for (Entry<Path, List<KeyExtent>> entry : assignmentFailures2.entrySet()) {
                assignmentFailures.get(entry.getKey()).addAll(entry.getValue());
                Integer fc = failureCount.get(entry.getKey());
                if (fc == null)
                    fc = 0;
                failureCount.put(entry.getKey(), fc + 1);
            }
            // remove map files that have no more key extents to assign
            Iterator<Entry<Path, List<KeyExtent>>> afIter = assignmentFailures.entrySet().iterator();
            while (afIter.hasNext()) {
                Entry<Path, List<KeyExtent>> entry = afIter.next();
                if (entry.getValue().size() == 0)
                    afIter.remove();
            }
            Set<Entry<Path, Integer>> failureIter = failureCount.entrySet();
            for (Entry<Path, Integer> entry : failureIter) {
                int retries = context.getConfiguration().getCount(Property.TSERV_BULK_RETRY);
                if (entry.getValue() > retries && assignmentFailures.get(entry.getKey()) != null) {
                    log.error("Map file {} failed more than {} times, giving up.", entry.getKey(), retries);
                    completeFailures.put(entry.getKey(), assignmentFailures.get(entry.getKey()));
                    assignmentFailures.remove(entry.getKey());
                }
            }
        }
        assignmentStats.assignmentsAbandoned(completeFailures);
        Set<Path> failedFailures = processFailures(completeFailures);
        assignmentStats.unrecoveredMapFiles(failedFailures);
        timer.stop(Timers.TOTAL);
        printReport(paths);
        return assignmentStats;
    } finally {
        if (client != null) {
            ServerClient.close(client);
        }
    }
}

Example 4 with TabletLocation

use of org.apache.accumulo.core.client.impl.TabletLocator.TabletLocation in project accumulo by apache.

the class BulkImporterTest method testFindOverlappingTablets.

@Test
public void testFindOverlappingTablets() throws Exception {
    MockTabletLocator locator = new MockTabletLocator();
    FileSystem fs = FileSystem.getLocal(CachedConfiguration.getInstance());
    ClientContext context = EasyMock.createMock(ClientContext.class);
    EasyMock.expect(context.getConfiguration()).andReturn(DefaultConfiguration.getInstance()).anyTimes();
    EasyMock.replay(context);
    String file = "target/testFile.rf";
    fs.delete(new Path(file), true);
    FileSKVWriter writer = FileOperations.getInstance().newWriterBuilder().forFile(file, fs, fs.getConf()).withTableConfiguration(context.getConfiguration()).build();
    writer.startDefaultLocalityGroup();
    Value empty = new Value(new byte[] {});
    writer.append(new Key("a", "cf", "cq"), empty);
    writer.append(new Key("a", "cf", "cq1"), empty);
    writer.append(new Key("a", "cf", "cq2"), empty);
    writer.append(new Key("a", "cf", "cq3"), empty);
    writer.append(new Key("a", "cf", "cq4"), empty);
    writer.append(new Key("a", "cf", "cq5"), empty);
    writer.append(new Key("d", "cf", "cq"), empty);
    writer.append(new Key("d", "cf", "cq1"), empty);
    writer.append(new Key("d", "cf", "cq2"), empty);
    writer.append(new Key("d", "cf", "cq3"), empty);
    writer.append(new Key("d", "cf", "cq4"), empty);
    writer.append(new Key("d", "cf", "cq5"), empty);
    writer.append(new Key("dd", "cf", "cq1"), empty);
    writer.append(new Key("ichabod", "cf", "cq"), empty);
    writer.append(new Key("icky", "cf", "cq1"), empty);
    writer.append(new Key("iffy", "cf", "cq2"), empty);
    writer.append(new Key("internal", "cf", "cq3"), empty);
    writer.append(new Key("is", "cf", "cq4"), empty);
    writer.append(new Key("iterator", "cf", "cq5"), empty);
    writer.append(new Key("xyzzy", "cf", "cq"), empty);
    writer.close();
    VolumeManager vm = VolumeManagerImpl.get(context.getConfiguration());
    List<TabletLocation> overlaps = BulkImporter.findOverlappingTablets(context, vm, locator, new Path(file));
    Assert.assertEquals(5, overlaps.size());
    Collections.sort(overlaps);
    Assert.assertEquals(new KeyExtent(tableId, new Text("a"), null), overlaps.get(0).tablet_extent);
    Assert.assertEquals(new KeyExtent(tableId, new Text("d"), new Text("cm")), overlaps.get(1).tablet_extent);
    Assert.assertEquals(new KeyExtent(tableId, new Text("dm"), new Text("d")), overlaps.get(2).tablet_extent);
    Assert.assertEquals(new KeyExtent(tableId, new Text("j"), new Text("i")), overlaps.get(3).tablet_extent);
    Assert.assertEquals(new KeyExtent(tableId, null, new Text("l")), overlaps.get(4).tablet_extent);
    List<TabletLocation> overlaps2 = BulkImporter.findOverlappingTablets(context, vm, locator, new Path(file), new KeyExtent(tableId, new Text("h"), new Text("b")));
    Assert.assertEquals(3, overlaps2.size());
    Assert.assertEquals(new KeyExtent(tableId, new Text("d"), new Text("cm")), overlaps2.get(0).tablet_extent);
    Assert.assertEquals(new KeyExtent(tableId, new Text("dm"), new Text("d")), overlaps2.get(1).tablet_extent);
    Assert.assertEquals(new KeyExtent(tableId, new Text("j"), new Text("i")), overlaps2.get(2).tablet_extent);
    Assert.assertEquals(locator.invalidated, 1);
}

Example 5 with TabletLocation

use of org.apache.accumulo.core.client.impl.TabletLocator.TabletLocation in project accumulo by apache.

the class TabletLocatorImplTest method createMetaCacheKE.

static TreeMap<KeyExtent, TabletLocation> createMetaCacheKE(Object... data) {
    TreeMap<KeyExtent, TabletLocation> mcke = new TreeMap<>();
    for (int i = 0; i < data.length; i += 2) {
        KeyExtent ke = (KeyExtent) data[i];
        String loc = (String) data[i + 1];
        mcke.put(ke, new TabletLocation(ke, loc, "1"));
    }
    return mcke;
}