Examples with RetryCounter org.apache.hadoop.hbase.util.RetryCounter used on opensource projects

Example 6 with RetryCounter

use of org.apache.hadoop.hbase.util.RetryCounter in project hbase by apache.

the class RecoverableZooKeeper method getChildren.

/**
   * getChildren is an idempotent operation. Retry before throwing exception
   * @return List of children znodes
   */
public List<String> getChildren(String path, Watcher watcher) throws KeeperException, InterruptedException {
    TraceScope traceScope = null;
    try {
        traceScope = Trace.startSpan("RecoverableZookeeper.getChildren");
        RetryCounter retryCounter = retryCounterFactory.create();
        while (true) {
            try {
                return checkZk().getChildren(path, watcher);
            } catch (KeeperException e) {
                switch(e.code()) {
                    case CONNECTIONLOSS:
                    case OPERATIONTIMEOUT:
                        retryOrThrow(retryCounter, e, "getChildren");
                        break;
                    default:
                        throw e;
                }
            }
            retryCounter.sleepUntilNextRetry();
        }
    } finally {
        if (traceScope != null)
            traceScope.close();
    }
}

Example 7 with RetryCounter

use of org.apache.hadoop.hbase.util.RetryCounter in project hbase by apache.

the class RecoverableZooKeeper method getData.

/**
   * getData is an idempotent operation. Retry before throwing exception
   * @return Data
   */
public byte[] getData(String path, Watcher watcher, Stat stat) throws KeeperException, InterruptedException {
    TraceScope traceScope = null;
    try {
        traceScope = Trace.startSpan("RecoverableZookeeper.getData");
        RetryCounter retryCounter = retryCounterFactory.create();
        while (true) {
            try {
                byte[] revData = checkZk().getData(path, watcher, stat);
                return removeMetaData(revData);
            } catch (KeeperException e) {
                switch(e.code()) {
                    case CONNECTIONLOSS:
                    case OPERATIONTIMEOUT:
                        retryOrThrow(retryCounter, e, "getData");
                        break;
                    default:
                        throw e;
                }
            }
            retryCounter.sleepUntilNextRetry();
        }
    } finally {
        if (traceScope != null)
            traceScope.close();
    }
}

Example 8 with RetryCounter

use of org.apache.hadoop.hbase.util.RetryCounter in project hbase by apache.

the class RecoverableZooKeeper method exists.

/**
   * exists is an idempotent operation. Retry before throwing exception
   * @return A Stat instance
   */
public Stat exists(String path, Watcher watcher) throws KeeperException, InterruptedException {
    TraceScope traceScope = null;
    try {
        traceScope = Trace.startSpan("RecoverableZookeeper.exists");
        RetryCounter retryCounter = retryCounterFactory.create();
        while (true) {
            try {
                return checkZk().exists(path, watcher);
            } catch (KeeperException e) {
                switch(e.code()) {
                    case CONNECTIONLOSS:
                    case OPERATIONTIMEOUT:
                        retryOrThrow(retryCounter, e, "exists");
                        break;
                    default:
                        throw e;
                }
            }
            retryCounter.sleepUntilNextRetry();
        }
    } finally {
        if (traceScope != null)
            traceScope.close();
    }
}

Example 9 with RetryCounter

use of org.apache.hadoop.hbase.util.RetryCounter in project hbase by apache.

the class ServerManager method isServerReachable.

/**
   * Check if a region server is reachable and has the expected start code
   */
public boolean isServerReachable(ServerName server) {
    if (server == null)
        throw new NullPointerException("Passed server is null");
    RetryCounter retryCounter = pingRetryCounterFactory.create();
    while (retryCounter.shouldRetry()) {
        try {
            HBaseRpcController controller = newRpcController();
            AdminService.BlockingInterface admin = getRsAdmin(server);
            if (admin != null) {
                ServerInfo info = ProtobufUtil.getServerInfo(controller, admin);
                return info != null && info.hasServerName() && server.getStartcode() == info.getServerName().getStartCode();
            }
        } catch (IOException ioe) {
            LOG.debug("Couldn't reach " + server + ", try=" + retryCounter.getAttemptTimes() + " of " + retryCounter.getMaxAttempts(), ioe);
            try {
                retryCounter.sleepUntilNextRetry();
            } catch (InterruptedException ie) {
                Thread.currentThread().interrupt();
            }
        }
    }
    return false;
}

Example 10 with RetryCounter

use of org.apache.hadoop.hbase.util.RetryCounter in project hbase by apache.

the class RegionReplicaFlushHandler method triggerFlushInPrimaryRegion.

void triggerFlushInPrimaryRegion(final HRegion region) throws IOException, RuntimeException {
    long pause = connection.getConfiguration().getLong(HConstants.HBASE_CLIENT_PAUSE, HConstants.DEFAULT_HBASE_CLIENT_PAUSE);
    int maxAttempts = getRetriesCount(connection.getConfiguration());
    RetryCounter counter = new RetryCounterFactory(maxAttempts, (int) pause).create();
    if (LOG.isDebugEnabled()) {
        LOG.debug("Attempting to do an RPC to the primary region replica " + ServerRegionReplicaUtil.getRegionInfoForDefaultReplica(region.getRegionInfo()).getEncodedName() + " of region " + region.getRegionInfo().getEncodedName() + " to trigger a flush");
    }
    while (!region.isClosing() && !region.isClosed() && !server.isAborted() && !server.isStopped()) {
        FlushRegionCallable flushCallable = new FlushRegionCallable(connection, rpcControllerFactory, RegionReplicaUtil.getRegionInfoForDefaultReplica(region.getRegionInfo()), true);
        // TODO: flushRegion() is a blocking call waiting for the flush to complete. Ideally we
        // do not have to wait for the whole flush here, just initiate it.
        FlushRegionResponse response = null;
        try {
            response = rpcRetryingCallerFactory.<FlushRegionResponse>newCaller().callWithRetries(flushCallable, this.operationTimeout);
        } catch (IOException ex) {
            if (ex instanceof TableNotFoundException || connection.isTableDisabled(region.getRegionInfo().getTable())) {
                return;
            }
            throw ex;
        }
        if (response.getFlushed()) {
            // a complete flush cycle or replay a region open event
            if (LOG.isDebugEnabled()) {
                LOG.debug("Successfully triggered a flush of primary region replica " + ServerRegionReplicaUtil.getRegionInfoForDefaultReplica(region.getRegionInfo()).getEncodedName() + " of region " + region.getRegionInfo().getEncodedName() + " Now waiting and blocking reads until observing a full flush cycle");
            }
            break;
        } else {
            if (response.hasWroteFlushWalMarker()) {
                if (response.getWroteFlushWalMarker()) {
                    if (LOG.isDebugEnabled()) {
                        LOG.debug("Successfully triggered an empty flush marker(memstore empty) of primary " + "region replica " + ServerRegionReplicaUtil.getRegionInfoForDefaultReplica(region.getRegionInfo()).getEncodedName() + " of region " + region.getRegionInfo().getEncodedName() + " Now waiting and " + "blocking reads until observing a flush marker");
                    }
                    break;
                } else {
                    // closing or already flushing. Retry flush again after some sleep.
                    if (!counter.shouldRetry()) {
                        throw new IOException("Cannot cause primary to flush or drop a wal marker after " + "retries. Failing opening of this region replica " + region.getRegionInfo().getEncodedName());
                    }
                }
            } else {
                // nothing to do. Are we dealing with an old server?
                LOG.warn("Was not able to trigger a flush from primary region due to old server version? " + "Continuing to open the secondary region replica: " + region.getRegionInfo().getEncodedName());
                region.setReadsEnabled(true);
                break;
            }
        }
        try {
            counter.sleepUntilNextRetry();
        } catch (InterruptedException e) {
            throw new InterruptedIOException(e.getMessage());
        }
    }
}