All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.netflix.astyanax.connectionpool.impl.AbstractHostPartitionConnectionPool Maven / Gradle / Ivy

There is a newer version: 3.10.2
Show newest version
/*******************************************************************************
 * Copyright 2011 Netflix
 * 
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 * 
 *   http://www.apache.org/licenses/LICENSE-2.0
 * 
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 ******************************************************************************/
package com.netflix.astyanax.connectionpool.impl;

import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.Comparator;
import java.util.List;
import java.util.Set;
import java.util.Map.Entry;

import org.cliffc.high_scale_lib.NonBlockingHashMap;

import com.google.common.collect.ImmutableList;
import com.google.common.collect.Lists;
import com.google.common.collect.Sets;
import com.netflix.astyanax.connectionpool.ConnectionFactory;
import com.netflix.astyanax.connectionpool.ConnectionPool;
import com.netflix.astyanax.connectionpool.ConnectionPoolConfiguration;
import com.netflix.astyanax.connectionpool.ConnectionPoolMonitor;
import com.netflix.astyanax.connectionpool.ExecuteWithFailover;
import com.netflix.astyanax.connectionpool.Host;
import com.netflix.astyanax.connectionpool.HostConnectionPool;
import com.netflix.astyanax.connectionpool.LatencyScoreStrategy.Listener;
import com.netflix.astyanax.connectionpool.Operation;
import com.netflix.astyanax.connectionpool.OperationResult;
import com.netflix.astyanax.connectionpool.TokenRange;
import com.netflix.astyanax.connectionpool.exceptions.ConnectionException;
import com.netflix.astyanax.connectionpool.exceptions.OperationException;
import com.netflix.astyanax.partitioner.Partitioner;
import com.netflix.astyanax.retry.RetryPolicy;

/**
 * Base for all connection pools that keep a separate pool of connections for
 * each host. 

* *

* Set of host connection pools
* The class maintains the set of all {@link HostConnectionPool}s for all hosts in the token ring using a non blocking hashmap and a {@link Topology} * The hashmap tracks the basic set of hosts and their corresponding connection pools. The topology is used to track the internal state of the token ring * for the cassandra cluster.
* The class uses these 2 structures to determine whether there has been a change to the system when a host joins or leaves the ring, * or even if an existing host just receives an update for the token ring partition that it owns. Hence the class can actively rebuild all the partitions * associated with each as and when there is a change.
* * See {@link #addHost(Host, boolean)} {@link #removeHost(Host, boolean)} {@link #setHosts(Collection)} for host changes and the corresponding trigger to {@link #rebuildPartitions()} * when this happens.
* * Note that when the connection pool is started it fetches the list of seed hosts from config (if any) and then inits it's data structures using these seed hosts.
* It also employs a listener to the latency score updates so that it can rebuild partitions as and when it receives updates *

* *

* Execute with failover
* * The class also provides a basic implementation of {@link #executeWithFailover(Operation, RetryPolicy)} by repeatedly consulting the {@link RetryPolicy}. Note that extending classes * must provide impl for the actual execute with failover. e.g the {@link RoundRobinConnectionPoolImpl} fails over to the next {@link HostConnectionPool} in the list. *

* * @author elandau * * @param * * @see {@link RoundRobinConnectionPoolImpl} {@link BagOfConnectionsConnectionPoolImpl} {@link TokenAwareConnectionPoolImpl} for details on impls that extend this class * @see {@link TokenPartitionedTopology} for how the internal topology structure is maintained across the set of hosts */ public abstract class AbstractHostPartitionConnectionPool implements ConnectionPool, SimpleHostConnectionPool.Listener { protected final NonBlockingHashMap> hosts; protected final ConnectionPoolConfiguration config; protected final ConnectionFactory factory; protected final ConnectionPoolMonitor monitor; protected final Topology topology; protected final Partitioner partitioner; /** * @param config * @param factory * @param monitor */ public AbstractHostPartitionConnectionPool(ConnectionPoolConfiguration config, ConnectionFactory factory, ConnectionPoolMonitor monitor) { this.config = config; this.factory = factory; this.monitor = monitor; this.hosts = new NonBlockingHashMap>(); this.topology = new TokenPartitionedTopology(config.getPartitioner(), config.getLatencyScoreStrategy()); this.partitioner = config.getPartitioner(); } /** * Starts the conn pool and resources associated with it */ @Override public void start() { ConnectionPoolMBeanManager.getInstance().registerMonitor(config.getName(), this); String seeds = config.getSeeds(); if (seeds != null && !seeds.isEmpty()) { setHosts(config.getSeedHosts()); } config.getLatencyScoreStrategy().start(new Listener() { @Override public void onUpdate() { rebuildPartitions(); } @Override public void onReset() { rebuildPartitions(); } }); } /** * Clean up resources associated with the conn pool */ @Override public void shutdown() { ConnectionPoolMBeanManager.getInstance().unregisterMonitor(config.getName(), this); for (Entry> pool : hosts.entrySet()) { pool.getValue().shutdown(); } config.getLatencyScoreStrategy().shutdown(); config.shutdown(); } protected HostConnectionPool newHostConnectionPool(Host host, ConnectionFactory factory, ConnectionPoolConfiguration config) { return new SimpleHostConnectionPool(host, factory, monitor, config, this); } /** * Host is marked as down * @param pool */ @Override public void onHostDown(HostConnectionPool pool) { topology.suspendPool(pool); } /** * Host is marked as up * @param pool */ @Override public void onHostUp(HostConnectionPool pool) { topology.resumePool(pool); } private static Comparator compareByStartToken = new Comparator() { @Override public int compare(TokenRange p1, TokenRange p2) { return p1.getStartToken().compareTo(p2.getStartToken()); } }; /** * Add host to the system. May need to rebuild the partition map of the system * @param host * @param refresh */ @Override public final synchronized boolean addHost(Host host, boolean refresh) { // Already exists if (hosts.containsKey(host)) { // Check to see if we are adding token ranges or if the token ranges changed // which will force a rebuild of the token topology Host existingHost = hosts.get(host).getHost(); if (existingHost.getTokenRanges().size() != host.getTokenRanges().size()) { existingHost.setTokenRanges(host.getTokenRanges()); return true; } ArrayList currentTokens = Lists.newArrayList(existingHost.getTokenRanges()); ArrayList newTokens = Lists.newArrayList(host.getTokenRanges()); Collections.sort(currentTokens, compareByStartToken); Collections.sort(newTokens, compareByStartToken); for (int i = 0; i < currentTokens.size(); i++) { if (!currentTokens.get(i).getStartToken().equals(newTokens.get(i).getStartToken()) || !currentTokens.get(i).getEndToken().equals(newTokens.get(i).getEndToken())) { return false; } } existingHost.setTokenRanges(host.getTokenRanges()); return true; } else { HostConnectionPool pool = newHostConnectionPool(host, factory, config); if (null == hosts.putIfAbsent(host, pool)) { try { monitor.onHostAdded(host, pool); if (refresh) { topology.addPool(pool); rebuildPartitions(); } pool.primeConnections(config.getInitConnsPerHost()); } catch (Exception e) { // Ignore, pool will have been marked down internally } return true; } else { return false; } } } /** * @return boolean */ @Override public boolean isHostUp(Host host) { HostConnectionPool pool = hosts.get(host); if (pool != null) { return !pool.isReconnecting(); } return false; } /** * @return boolean */ @Override public boolean hasHost(Host host) { return hosts.containsKey(host); } /** * list of all active pools * @return {@link List} */ @Override public List> getActivePools() { return ImmutableList.copyOf(topology.getAllPools().getPools()); } /** * @return {@link List} */ @Override public List> getPools() { return ImmutableList.copyOf(hosts.values()); } /** * Remove host from the system. Shuts down pool associated with the host and rebuilds partition map * @param host * @param refresh */ @Override public synchronized boolean removeHost(Host host, boolean refresh) { HostConnectionPool pool = hosts.remove(host); if (pool != null) { topology.removePool(pool); rebuildPartitions(); monitor.onHostRemoved(host); pool.shutdown(); return true; } else { return false; } } /** * @param host * @return {@link HostConnectionPool} */ @Override public HostConnectionPool getHostPool(Host host) { return hosts.get(host); } /** * @param ring */ @Override public synchronized void setHosts(Collection ring) { // Temporary list of hosts to remove. Any host not in the new ring will // be removed Set hostsToRemove = Sets.newHashSet(hosts.keySet()); // Add new hosts. boolean changed = false; for (Host host : ring) { if (addHost(host, false)) changed = true; hostsToRemove.remove(host); } // Remove any hosts that are no longer in the ring for (Host host : hostsToRemove) { removeHost(host, false); changed = true; } if (changed) { topology.setPools(hosts.values()); rebuildPartitions(); } } /** * Executes the operation using failover and retry strategy * @param op * @param retry * @return {@link OperationResult} */ @Override public OperationResult executeWithFailover(Operation op, RetryPolicy retry) throws ConnectionException { retry.begin(); ConnectionException lastException = null; do { try { OperationResult result = newExecuteWithFailover(op).tryOperation(op); retry.success(); return result; } catch (OperationException e) { retry.failure(e); throw e; } catch (ConnectionException e) { lastException = e; } } while (retry.allowRetry()); retry.failure(lastException); throw lastException; } /** * Return a new failover context. The context captures the connection pool * state and implements the necessary failover logic. * * @param * @return * @throws ConnectionException */ protected abstract ExecuteWithFailover newExecuteWithFailover(Operation op) throws ConnectionException; /** * Called every time a host is added, removed or is marked as down */ protected void rebuildPartitions() { topology.refresh(); } /** * @return {@link Topology} */ public Topology getTopology() { return topology; } /** * @return {@link Partitioner} */ public Partitioner getPartitioner() { return this.partitioner; } }




© 2015 - 2024 Weber Informatics LLC | Privacy Policy