@Override
public void onStop() {
if (conn != null) {
conn.close();
conn = null;
producer.close();
consumer.close();
}
}
@Override
public void run() {
logger.info("begin to consume tweets");
controller.running();
while (true) {
susSem.acquireUninterruptibly();
try {
for (Entry<String, List<Status>> topicData : consumer.nextStatus().entrySet()) {
try {
// 在高速转发的情况下，这样可以减轻写入到kafka中的状态数据
Map<String, TimeSeriesUpdateState> states =
new HashMap<String, TimeSeriesUpdateState>();
List<TimeSeriesUpdateState> segSignal = new ArrayList<TimeSeriesUpdateState>();
for (Status cur : topicData.getValue()) {
tweetDao.putTweet(cur);
if (cur.getRetweetedStatus() != null) {
// rand.nextFloat() < 0.01 &&
if (tweetDao.getStatusByMid(cur.getRetweetedStatus().getMid()) == null) {
tweetDao.putTweet(cur.getRetweetedStatus());
}
tweetDao.putRtweet(cur);
for (TimeSeriesUpdateState state : tweetDao.updateRtTimeSeries(cur)) {
states.put(state.getMid(), state);
}
} else {
// maybe it is the first tweet or an indication
// for end of monitoring
if (cur.getMid() != null) {
SegState state = segDao.getSegState(cur.getMid());
if (state != null) {
long updateDate = DateUtil.roundByHour(System.currentTimeMillis());
segSignal.add(new TimeSeriesUpdateState(state.mid, updateDate, true));
List<String> mids = tweetDao.getRtMids(state.mid);
for (String rtMid : mids) {
segSignal.add(new TimeSeriesUpdateState(rtMid, updateDate, true));
}
}
}
}
}
for (TimeSeriesUpdateState state : segSignal) {
states.put(state.mid, state);
}
for (TimeSeriesUpdateState state : states.values()) {
logger.info("update time series " + state);
producer.storeTsUpdateState(state);
}
} catch (Exception ex) {
ex.printStackTrace();
}
}
} finally {
susSem.release();
}
}
}
public void init(String kafkaServers, String dbServers, boolean restart) throws Exception {
logger.info("connecting to cassandra");
conn = new CassandraConn();
conn.connect(dbServers);
logger.info("create TweetDao");
tweetDao = new TweetDao(conn);
segDao = new SegStateDao(conn);
logger.info("connecting to kafka");
consumer = new TweetConsumer();
consumer.open(
Arrays.asList(KafkaTopics.RETWEET_TOPIC),
KafkaTopics.RTSERIES_GROUP,
kafkaServers,
restart);
producer = new TweetKafkaProducer(KafkaProducerFactory.createProducer(kafkaServers));
MetricBasedPerfProfile.registerServer(controller);
}
