kinesis firehose 转换kinesis流数据为parquet 格式存储到S3上

一、beanstalk部署
1、创建两个存储桶
sengled-bucket-reliabilityops-backuptemp
sengled-bucket-reliabilityops-kinesis

2、配置glue
1）创建数据库：只填写名称即可，不需要填写位置信息
2）配置爬网程序：主要填写名称、数据库、s3源数据存放桶路径（需要根据谁定义表结构，就输入谁的存放路径）
3）运行爬网程序自动生成“数据目录”的表

3、配置glue任务（主要是脚本编写）
import sys
from awsglue.transforms import *
from pyspark.context import SparkContext
from awsglue.context import GlueContext
from awsglue.utils import getResolvedOptions
from pyspark.sql.types import TimestampType,DateType
from awsglue.job import Job
import boto3
import datetime
import time
import logging

logging.basicConfig()
logger = logging.getLogger()
logger.setLevel(logging.INFO)

## @params: [JOB_NAME]
args = getResolvedOptions(sys.argv, ['JOB_NAME'])

sc = SparkContext()
glueContext = GlueContext(sc)
spark = glueContext.spark_session
job = Job(glueContext)
job.init(args['JOB_NAME'], args)

start_hour_before = 1
end_hour_before = 1

source_bucket_name = "backuptemp"
target_bucket_name = "kinesis"
target_prefix_name = "alertlog"
delimiter = "/"
default_region = "us-west-2"
crawler_name = "clean_alertlog"

sns_arn = "arn:aws:sns:us-east-1:67789:logprocess"
client = boto3.client('s3')

def delete_object(bucket_name,key_name):
    try:
            response = client.delete_object(Bucket=bucket_name,Key=key_name)
    except Exception as e:
            print str(e)
                #email_alert("error when delete_object %s/%s" % (bucket_name, key_name))
                
def email_alert(message):
    sns_client = boto3.client('sns',region_name = default_region)
    response = sns_client.publish(
        TopicArn=sns_arn,
        Message=message,
        Subject="glue %s process error" % args['JOB_NAME']
    )
    logger.error(response)

def aggragate_files(date,key_name):
    logger.info("start aggragate %s, time is %s." % (key_name, time.strftime('%Y-%m-%d %H:%M:%S',time.localtime(time.time()))))
    if key_name == "":
        return

    try:
        dataframe = spark.read.json("s3://%s/%s" % (source_bucket_name,key_name))
        print("dataframe.....................",dataframe)
        dataframe.write.parquet("s3://%s/%s/dt=%s" % (target_bucket_name, target_prefix_name, date), mode="append")
        logger.info("finish aggragate %s, time is %s." % (key_name, time.strftime('%Y-%m-%d %H:%M:%S',time.localtime(time.time()))))
    except Exception as e:
        #email
        #email_alert("error when aggragate %s/%s/%s: %s." % (key_name, date, hour, str(e)))
        print str(e)
    else:
            delete_object(source_bucket_name,key_name)
                
def main():
    s3 = boto3.resource('s3')
    process_slot = datetime.datetime.now() - datetime.timedelta(days=start_hour_before)
    bucket = s3.Bucket(source_bucket_name)
    dt = process_slot.strftime("%Y-%m-%d")
    for obj in bucket.objects.all():
        aggragate_files(dt,obj.key)
        
    glue_client = boto3.client('glue', region_name = default_region)
    response = glue_client.start_crawler(Name=crawler_name)

main()

####commit job
job.commit()

4、创建glue触发器

5、在beanstalk上部署java版本的kinesis流数据备份服务

二、kinesis firehost 部署
1、创建两个s3存储桶
sengled-bucket-reliabilityops-gluetable
sengled-bucket-reliabilityops-gluedatabase

2、配置glue
1）创建数据库：只填写名称即可，不需要填写位置信息
2）配置爬网程序：主要填写名称、数据库、s3源数据存放桶路径（sengled-bucket-reliabilityops-gluetable）勾选Create a single schema for each S3 path
3）运行爬网程序自动生成“数据目录”的表

3、配置firehost
填写名称（sengled-firehose-reliabilityOps-falcon）
选择源数据：kinesis
lambda：由于源数据就是json格式，所以不需要转换成json格式，这里设置：Disabled
格式转换：启用，默认Apache Parquet格式
glue选择区域、数据库名称、表名称、glue表的版本
目标S3：sengled-bucket-reliabilityops-gluedatabase
备份S3：sengled-bucket-reliabilityops-gluetable （用于每小时运行爬网程序更新glue表结构）

4、新建一个glue 任务用于定期删除备份数据，备份数据在这里只做定时运行爬网程序使用。