I01-通过查询资料库方式来监控Informatica调度情况

--登陆INFA资料库，运行下面的SQL

--想要更加个性化查询的话注意看SQL倒数第二第三行的备注

SELECT
    RUN_DATE,
    START_TIME ,
    END_TIME,
    FOLIDER ,
    WORKFLOW,
    WORKLET_LVL3,
    WORKLET_LVL2,
    WORKLET_LVL1,
    SESSION_NAME,
    MAPPING_NAME,
    RUN_STATUS,
    SUSSESSFUL_SOURC_ROWS,
    FAILED_SOURCE_ROWS,
    SUCCCESSFUL_ROWS,
    FAILED_ROWS,
    RUN_ERR_CODE,
    RUN_ERR_MSG,
    FIRST_ERROR_CODE,
    FIRST_ERROR_MSG,
    LAST_ERROR_CODE,
    LAST_ERROR,
    SESSION_LOG_FILE,
    BAD_FILE_LOCATION
FROM (SELECT
        TRUNC(T1.START_TIME) AS RUN_DATE, --跑数日期
        T1.START_TIME AS START_TIME, --开始时间
        T1.END_TIME AS END_TIME, --结束时间
        T.SUBJECT_AREA AS FOLIDER, --FOLDER
        T.WORKFLOW_NAME AS WORKFLOW, --WORKFLOW
        T4.INSTANCE_NAME AS WORKLET_LVL3, --大爷级WORKLET
        T3.INSTANCE_NAME AS WORKLET_LVL2, --爷爷级WORKLET
        T2.INSTANCE_NAME AS WORKLET_LVL1, --父级WORKLET
        T1.INSTANCE_NAME AS SESSION_NAME, --SESSION
        decode(T1.RUN_STATUS_CODE,
                1,'Succeeded',
                2,'Disabled',
                3,'Failed',
                4,'Stopped',
                5,'Aborted',
                6,'Running',
                7,'Suspending',
                8,'Suspended',
                9,'Stopping',
                10,'Aborting',
                11,'Waiting',
                12,'Scheduled',
                13,'Unscheduled',
                14,'Unknown',
                15,'Terminated'
            ) as RUN_STATUS, --运行状态
        T1.RUN_ERR_CODE RUN_ERR_CODE, --错误代码
        T1.RUN_ERR_MSG RUN_ERR_MSG, --错误信息
        T.FIRST_ERROR_CODE, --第一次发生错误的代码
        T.FIRST_ERROR_MSG, --第一次发生错误的信息
        T.LAST_ERROR_CODE, --最后发生错误的代码
        T.LAST_ERROR, --最后发生错误的信息
        T.SESSION_LOG_FILE, --SESSION LOG路径
        T.BAD_FILE_LOCATION, --BAD_FILE路径
        T.MAPPING_NAME MAPPING_NAME, --MAPPING
        T.SUCCESSFUL_SOURCE_ROWS SUSSESSFUL_SOURC_ROWS, --读取源成功条数
        T.FAILED_SOURCE_ROWS FAILED_SOURCE_ROWS, --读取源失败条数
        T.SUCCESSFUL_ROWS SUCCCESSFUL_ROWS, --插入目标成功条数
        T.FAILED_ROWS FAILED_ROWS --插入目标失败条数
    FROM REP_SESS_LOG T
INNER JOIN REP_TASK_INST_RUN T1 --返回session状态信息
    ON T.SUBJECT_ID = T1.SUBJECT_ID
        AND T.WORKFLOW_ID = T1.WORKFLOW_ID
        AND T.WORKFLOW_RUN_ID = T1.WORKFLOW_RUN_ID
        AND T.WORKLET_RUN_ID = T1.WORKLET_RUN_ID
        AND T.SESSION_ID = T1.TASK_ID
        AND T.INSTANCE_ID = T1.INSTANCE_ID
        AND T1.TASK_TYPE_NAME = 'Session'
 
LEFT JOIN REP_TASK_INST_RUN T2 --返回父级WKL
    ON T1.SUBJECT_ID = T2.SUBJECT_ID
        AND T1.WORKFLOW_ID = T2.WORKFLOW_ID
        AND T1.WORKFLOW_RUN_ID = T2.WORKFLOW_RUN_ID
        AND T1.WORKLET_RUN_ID = T2.CHILD_RUN_ID
        AND T2.TASK_TYPE_NAME = 'Worklet'
 
LEFT JOIN REP_TASK_INST_RUN T3 --返回爷爷级WKL，到这一级其实就可以看到session是属于哪个电厂的了
    ON T2.SUBJECT_ID = T3.SUBJECT_ID
        AND T2.WORKFLOW_ID = T3.WORKFLOW_ID
        AND T2.WORKFLOW_RUN_ID = T3.WORKFLOW_RUN_ID
        AND T2.WORKLET_RUN_ID = T3.CHILD_RUN_ID
        AND T3.TASK_TYPE_NAME = 'Worklet'
 
LEFT JOIN REP_TASK_INST_RUN T4 --返回老大爷级WKL
    ON T3.SUBJECT_ID = T4.SUBJECT_ID
        AND T3.WORKFLOW_ID = T4.WORKFLOW_ID
        AND T3.WORKFLOW_RUN_ID = T4.WORKFLOW_RUN_ID
        AND T3.WORKLET_RUN_ID = T4.CHILD_RUN_ID
        AND T4.TASK_TYPE_NAME = 'Worklet'
)
 
WHERE
START_TIME > TRUNC(SYSDATE-1) ------筛选时间
AND RUN_STATUS = 'Failed' --session运行状态的筛选，若只想看失败的session信息，则令RUN_STATUS = 'Failed'；否则去掉该条件,此时可以看到所有session运行状态
 
ORDER BY RUN_DATE DESC, WORKFLOW, WORKLET_LVL3, WORKLET_LVL2, WORKLET_LVL1, SESSION_NAME --排序以更好的形式展示出来

查询得到的结果示例如下：

拿第一条举例，由RUN_STATUS这个字段内容（Failed）可以看出，ETL0_ODS这个folder下的WKF_ETL0这个workflow下的WKL_FMIS_GJXNY worklet下的WKL_T07 worklet 下的WKL_FMIS_APPLSYS worklet下的s_m_ods_fmis_applsys_fnd_flex_value_sets这个session调度出错了，出错原因是无法连接上数据库。

如果想让该SQL每天自动运行，然后将查数结果发送到邮箱的话，我的做法需要下面几个步骤：

（1）将该sql保存到Informatica服务器上的一个文件中（base_sql.txt），有可能需要去掉上面SQL里的中文注释。

（2）Informatica服务器上写一个shell脚本（workflow_dispatch.sh），脚本中调用sqluldr2命令来远程到Informatica资料库上执行base_sql.txt文件中的SQL语句，结果保存到一个文件中。命令如下：

 sqluldr2 user=${conn_str} sql=${py_file_dir}${sql_file} fast=yes field="," head=yes file=${result_dir}${result_file} escape='\' escf=0x0a esct=n

其中escape='\' escf=0x0a esct=n等参数设定是为了去掉sql执行结果中有的数据里面存在换行。

（3）执行完第二步之后，查数结果已经保存到某个文件中（ sqluldr2命令中file参数指定的文件）了，接下来只需在shell中调用相应的邮件服务命令将结果文件发送到邮件即可。至于Informatica服务器上的邮件服务，使用msmtp+mutt或者写python脚本或者其他的都行，我采用的python脚本的方式（sendmail.py）。若是想使用msmtp+mutt的方式，可参考以下链接进行配置：http://www.cnblogs.com/suhaha/p/8655033.html。

（4）虽然存放查数结果的文件占用空间不多，也应该在shell中定时删除太久之前的文件，比如每次都固定删除一个月之前的文件。

完整的workflow_dispatch.sh脚本示例如下：

#!/bin/bash
 
check_date=$(date +%Y%m%d_%H%M)
 
base_dir="/data/ODS_TO_EDW_DATACHECK/"
result_dir=${base_dir}"result_file/"
py_file_dir=${base_dir}"py_file/"
 
#conn_str="repdb01/infa@10.xxx.xxx.182:1521/test"    #DEV environment
conn_str="repdb/oracle@10.xxx.xxx.165:1521/appdb"    #PRD environment
 
result_file="dispatch_result_"${check_date}".csv"
sql_file="base_sql.txt"
tar_file="dispatch_result_"${check_date}".tar.gz"
mail_file="sendmail.py"
 
echo "execute sqluldr2..."
echo -e "\n"
 
sqluldr2 user=${conn_str} sql=${py_file_dir}${sql_file} fast=yes field="," head=yes file=${result_dir}${result_file} escape='\' escf=0x0a esct=n
 
if [ $? -eq  ]; then
    echo -e "\n"
    echo "sqluldr2 run successful!"
    echo -e "\n"
 
    echo "check the count of error session..."
    cd ${result_dir}
    cnt=`cat ${result_file} | wc -l`
    cnt=`expr ${cnt} - `
    echo "[cnt of error sessions ]: "${cnt}
    echo -e "\n"
 
    #echo "compress the result file into : "${tar_file}
    #tar -czvf ${tar_file} ${result_file} && rm -rf ${result_file}
 
    echo -e "\n"
    echo "Send mail..."
    if [ ${cnt} -gt  ]; then
        python ${py_file_dir}${mail_file} "Result of Informatica Dispatch (PRD)" \
            "Infamatica Dispatch finish. "${cnt}" error sessions. Check the attachment for more details." ${result_dir}${result_file} $@
    else
        python ${py_file_dir}${mail_file} "Result of Informatica Dispatch (PRD)" "Infamatica Dispatch finish. No error." ${result_dir}${result_file} $@
    fi
 
else
    echo "sqluldr2 run failed!"
    exit
fi

完整的sendmail.py脚本示例如下：

#!/usr/bin/python
# -*- coding: UTF-8 -*-
 
import sys
import smtplib
from email.mime.text import MIMEText
from email.mime.multipart import MIMEMultipart
from email.header import Header
from email.utils import formataddr
 
"""
        该脚本调用格式: python /data/ODS_TO_EDW_DATACHECK/py_file/sendmail.py  +theme +body +attachment +recipient1 +recipient2  + ... +recipients
        调用示例:python sendmail.py  "Result of DataCheck" "23333..This is body..."  /data/ODS_TO_EDW_DATACHECK/py_file/sendmail.py 983523649@qq.com
 
        备注:添加的附件必须是全路径的，比如:/data/ODS_TO_EDW_DATACHECK/py_file/sendmail.py
"""
 
def sendmail( theme, body, attachment, recipients ):
 
    mail_host="smtp.163.com"
    mail_user="182xxxx7782@163.com"
    mail_pass="wysxxx9527"
 
    subject = theme
    text = body
    file = attachment
    receivers = recipients
    sender = "182xxx7782@163.com"
 
    message = MIMEMultipart()
    message["From"] = formataddr(["INFA SERVER", sender])
    message["To"] =  Header("ETL Engineer", "utf-8")
    message["Subject"] = Header(subject, "utf-8")
    message.attach(MIMEText(text, "plain", "utf-8"))
    name = file[ file.rindex(r"/")+1 : len(file) ]
    att = MIMEText(open(file, "r").read(), "base64", "utf-8")
    att["Content-Type"] = "application/octet-stream"
    att["Content-Disposition"] = 'attachment; filename=' + '"' + name + '"'
    message.attach(att)
 
    try:
        smtpObj = smtplib.SMTP()
        smtpObj.connect(mail_host, 25)
        smtpObj.login(mail_user,mail_pass)
        smtpObj.sendmail(sender, receivers, message.as_string())
        print "\nSend Mail Successfully!"
    except smtplib.SMTPException:
        print "\nError: Send Mail Failed!"
        print str(smtplib.SMTPException)
 
#*********************************************************************************************
if __name__ == '__main__':
 
    theme = sys.argv[1]
    body = sys.argv[2]
    attachment = sys.argv[3]
    recipients = sys.argv[4: len(sys.argv)]
 
    print "theme: ", theme
    print "body: ", body
    print "attachment: ", attachment
    print "recipients: ", recipients
 
    sendmail( theme, body, attachment, recipients )

（5）在Informatica中新建一个Task，Task中调用shell脚本。如下图。然后在INFA每日调度的最后一个session或者最后一个workflow末尾加上该Task即可。

最后，当然INFA自身也能配置邮件服务，并且配置之后它的报错也能详细到session级别，如下图。说实话我没用过。

不过在项目的实施过程中，开发的session数量一般都是上几百个的，每个session都配置的话会比较麻烦；再者，假设真的每个session都配置了，然后每个session跑成功或者失败了都给你发一封邮件，一般INFA的调度还都在半夜，会疯掉的~

我之所采用上面描述的方法是因为：项目开发之初压根没考虑到调度的监控问题，都项目后期了前台数据对不上之后往前溯源了才发现是某个session挂掉好久了，因此才想起配置这么个东西。

另外，我觉得这种方式比INFA自身的邮件设置方式要相对好一点吧~INFA调度完成之后统一查询，速度快，效率也高，最重要的是每天只给你发一封邮件。。。