use of org.I0Itec.zkclient.exception.ZkException in project otter by alibaba.
the class ZooKeeperx method connect.
@Override
public void connect(Watcher watcher) {
ReflectionUtils.makeAccessible(zookeeperLockField);
ReflectionUtils.makeAccessible(zookeeperFiled);
Lock _zookeeperLock = (ReentrantLock) ReflectionUtils.getField(zookeeperLockField, this);
ZooKeeper _zk = (ZooKeeper) ReflectionUtils.getField(zookeeperFiled, this);
_zookeeperLock.lock();
try {
if (_zk != null) {
throw new IllegalStateException("zk client has already been started");
}
String zkServers = _serversList.get(0);
try {
logger.debug("Creating new ZookKeeper instance to connect to " + zkServers + ".");
_zk = new ZooKeeper(zkServers, _sessionTimeOut, watcher);
configMutliCluster(_zk);
ReflectionUtils.setField(zookeeperFiled, this, _zk);
} catch (IOException e) {
throw new ZkException("Unable to connect to " + zkServers, e);
}
} finally {
_zookeeperLock.unlock();
}
}
use of org.I0Itec.zkclient.exception.ZkException in project otter by alibaba.
the class ZooKeeperx method configMutliCluster.
// ===============================
public void configMutliCluster(ZooKeeper zk) {
if (_serversList.size() == 1) {
return;
}
String cluster1 = _serversList.get(0);
try {
if (_serversList.size() > 1) {
// 强制的声明accessible
ReflectionUtils.makeAccessible(clientCnxnField);
ReflectionUtils.makeAccessible(hostProviderField);
ReflectionUtils.makeAccessible(serverAddressesField);
// 添加第二组集群列表
for (int i = 1; i < _serversList.size(); i++) {
String cluster = _serversList.get(i);
// 强制获取zk中的地址信息
ClientCnxn cnxn = (ClientCnxn) ReflectionUtils.getField(clientCnxnField, zk);
HostProvider hostProvider = (HostProvider) ReflectionUtils.getField(hostProviderField, cnxn);
List<InetSocketAddress> serverAddrs = (List<InetSocketAddress>) ReflectionUtils.getField(serverAddressesField, hostProvider);
// 添加第二组集群列表
serverAddrs.addAll(new ConnectStringParser(cluster).getServerAddresses());
}
}
} catch (Exception e) {
try {
if (zk != null) {
zk.close();
}
} catch (InterruptedException ie) {
// ignore interrupt
}
throw new ZkException("zookeeper_create_error, serveraddrs=" + cluster1, e);
}
}
use of org.I0Itec.zkclient.exception.ZkException in project otter by alibaba.
the class ExtractZooKeeperArbitrateEvent method await.
// private TerminArbitrateEvent terminEvent;
/**
* <pre>
* 算法:
* 1. 检查当前的Permit,阻塞等待其授权(解决Channel的pause状态处理)
* 2. 开始阻塞获取符合条件的processId
* 3. 检查当前的即时Permit状态 (在阻塞获取processId过程会出现一些error信号,process节点会被删除)
* 4. 获取Select传递的EventData数据,添加next node信息后直接返回
* </pre>
*
* @return
*/
public EtlEventData await(Long pipelineId) throws InterruptedException {
Assert.notNull(pipelineId);
PermitMonitor permitMonitor = ArbitrateFactory.getInstance(pipelineId, PermitMonitor.class);
// 阻塞等待授权
permitMonitor.waitForPermit();
ExtractStageListener extractStageListener = ArbitrateFactory.getInstance(pipelineId, ExtractStageListener.class);
// 符合条件的processId
Long processId = extractStageListener.waitForProcess();
ChannelStatus status = permitMonitor.getChannelPermit();
if (status.isStart()) {
// 即时查询一下当前的状态,状态随时可能会变
// 根据pipelineId+processId构造对应的path
String path = StagePathUtils.getSelectStage(pipelineId, processId);
try {
byte[] data = zookeeper.readData(path);
EtlEventData eventData = JsonUtils.unmarshalFromByte(data, EtlEventData.class);
// 获取下一个处理节点信息
Node node = LoadBalanceFactory.getNextTransformNode(pipelineId);
if (node == null) {
// terminEvent.single(termin);
throw new ArbitrateException("Extract_single", "no next node");
} else {
eventData.setNextNid(node.getId());
// 只有这一条路返回
return eventData;
}
} catch (ZkNoNodeException e) {
logger.error("pipeline[{}] processId[{}] is invalid , retry again", pipelineId, processId);
// /出现节点不存在,说明出现了error情况,递归调用重新获取一次
return await(pipelineId);
} catch (ZkException e) {
throw new ArbitrateException("Extract_await", e.getMessage(), e);
}
} else {
logger.warn("pipelineId[{}] extract ignore processId[{}] by status[{}]", new Object[] { pipelineId, processId, status });
// 释放下processId,因为load是等待processId最小值完成Tranform才继续,如果这里不释放,会一直卡死等待
String path = StagePathUtils.getProcess(pipelineId, processId);
zookeeper.delete(path);
// 递归调用
return await(pipelineId);
}
}
use of org.I0Itec.zkclient.exception.ZkException in project otter by alibaba.
the class SelectProcessListener method recovery.
/**
* 尝试载入一下上一次未使用的processId,可能发生mainstem切换,新的S模块需要感知前S模块已创建但未使用的process,不然就是一个死锁。而针对已经使用的processId会由e/t/l节点进行处理
*/
private void recovery(Long pipelineId) {
List<Long> currentProcessIds = processMonitor.getCurrentProcessIds(false);
for (Long processId : currentProcessIds) {
String path = StagePathUtils.getProcess(pipelineId, processId);
try {
byte[] bytes = zookeeper.readData(path);
ProcessNodeEventData nodeData = JsonUtils.unmarshalFromByte(bytes, ProcessNodeEventData.class);
if (nodeData.getStatus().isUnUsed()) {
// 加入未使用的processId
addReply(processId);
}
} catch (ZkException e) {
logger.error("recovery error!", e);
}
}
}
use of org.I0Itec.zkclient.exception.ZkException in project otter by alibaba.
the class LoadZooKeeperArbitrateEvent method await.
// private Map<Long, DistributedLock> locks = new ConcurrentHashMap<Long,
// DistributedLock>();
/**
* <pre>
* 算法:
* 1. 检查当前的Permit,阻塞等待其授权(解决Channel的pause状态处理)
* 2. 开始阻塞获取符合条件的processId
* 3. 检查当前的即时Permit状态 (在阻塞获取processId过程会出现一些error信号,process节点会被删除)
* 4. 获取Select传递的EventData数据,添加next node信息后直接返回
* </pre>
*/
public EtlEventData await(Long pipelineId) throws InterruptedException {
Assert.notNull(pipelineId);
PermitMonitor permitMonitor = ArbitrateFactory.getInstance(pipelineId, PermitMonitor.class);
// 阻塞等待授权
permitMonitor.waitForPermit();
LoadStageListener loadStageListener = ArbitrateFactory.getInstance(pipelineId, LoadStageListener.class);
// 符合条件的processId
Long processId = loadStageListener.waitForProcess();
// DistributedLock lock = getLock(pipelineId);
try {
// 使用锁的理由:
// 1. 针对双向同步时,其中一个方向出现了异常,需要发起另一端的关闭,此时对方正好在执行某个process的load
// 2. 单向同步时,如果出现node节点异常,此时正常的节点正在执行某个process的load
// 为避免因load无法中端引起的数据重复录入,所以针对load阶段添加分布式锁。在有process
// load过程中不允许进行pipeline关闭操作
// lock.lock();
ChannelStatus status = permitMonitor.getChannelPermit();
if (status.isStart()) {
// 即时查询一下当前的状态,状态随时可能会变
// 根据pipelineId+processId构造对应的path
String path = StagePathUtils.getTransformStage(pipelineId, processId);
try {
byte[] data = zookeeper.readData(path);
// 反序列化并返回
return JsonUtils.unmarshalFromByte(data, EtlEventData.class);
} catch (ZkNoNodeException e) {
logger.error("pipeline[{}] processId[{}] is invalid , retry again", pipelineId, processId);
// /出现节点不存在,说明出现了error情况,递归调用重新获取一次
return await(pipelineId);
} catch (ZkException e) {
throw e;
}
} else {
logger.warn("pipelineId[{}] load ignore processId[{}] by status[{}]", new Object[] { pipelineId, processId, status });
// try {
// lock.unlock();// 出现任何异常解除lock
// } catch (KeeperException e) {
// // ignore
// }
// 释放下processId,因为load是等待processId最小值完成Tranform才继续,如果这里不释放,会一直卡死等待
String path = StagePathUtils.getProcess(pipelineId, processId);
zookeeper.delete(path);
// 出现rollback情况,递归调用重新获取一次,当前的processId可丢弃
return await(pipelineId);
}
} catch (InterruptedException e) {
throw e;
} catch (Exception e) {
throw new ArbitrateException(e);
}
}
Aggregations