Open-Falcon 监控系统监控 MySQL/Redis/MongoDB 状态监控
背景：
Open-Falcon 是小米运维部开源的一款互联网企业级监控系统解决方案，具体的安装和使用说明请见官网：http://open-falcon.org/，是一款比较全的监控。而且提供各种API，只需要把数据按照规定给出就能出图，以及报警、集群支持等等。
监控：
1) MySQL 收集信息脚本(mysql_monitor.py)
#!/bin/env python
# -*- encoding: utf-8 -*-
from __future__ import division
import MySQLdb
import datetime
import time
import os
import sys
import fileinput
import requests
import json
import re
class MySQLMonitorInfo():
def __init__(self,host,port,user,password):
self.host = host
self.port = port
self.user = user
self.password = password
def stat_info(self):
try:
m = MySQLdb.connect(host=self.host,user=self.user,passwd=self.password,port=self.port,charset='utf8')
query = "SHOW GLOBAL STATUS"
cursor = m.cursor()
cursor.execute(query)
Str_string = cursor.fetchall()
Status_dict = {}
for Str_key,Str_value in Str_string:
Status_dict[Str_key] = Str_value
cursor.close()
m.close()
return Status_dict
except Exception, e:
print (datetime.datetime.now()).strftime("%Y-%m-%d %H:%M:%S")
print e
Status_dict = {}
return Status_dict
def engine_info(self):
try:
m = MySQLdb.connect(host=self.host,user=self.user,passwd=self.password,port=self.port,charset='utf8')
_engine_regex = re.compile(ur'(History list length) ([0-9]+\.?[0-9]*)\n')
query = "SHOW ENGINE INNODB STATUS"
cursor = m.cursor()
cursor.execute(query)
Str_string = cursor.fetchone()
a,b,c = Str_string
cursor.close()
m.close()
return dict(_engine_regex.findall(c))
except Exception, e:
print (datetime.datetime.now()).strftime("%Y-%m-%d %H:%M:%S")
print e
return dict(History_list_length=0)
if __name__ == '__main__':
open_falcon_api = 'http://192.168.200.86:1988/v1/push'
db_list= []
for line in fileinput.input():
db_list.append(line.strip())
for db_info in db_list:
# host,port,user,password,endpoint,metric = db_info.split(',')
host,port,user,password,endpoint = db_info.split(',')
timestamp = int(time.time())
step = 60
# tags = "port=%s" %port
tags = ""
conn = MySQLMonitorInfo(host,int(port),user,password)
stat_info = conn.stat_info()
engine_info = conn.engine_info()
mysql_stat_list = []
monitor_keys = [
('Com_select','COUNTER'),
('Qcache_hits','COUNTER'),
('Com_insert','COUNTER'),
('Com_update','COUNTER'),
('Com_delete','COUNTER'),
('Com_replace','COUNTER'),
('MySQL_QPS','COUNTER'),
('MySQL_TPS','COUNTER'),
('ReadWrite_ratio','GAUGE'),
('Innodb_buffer_pool_read_requests','COUNTER'),
('Innodb_buffer_pool_reads','COUNTER'),
('Innodb_buffer_read_hit_ratio','GAUGE'),
('Innodb_buffer_pool_pages_flushed','COUNTER'),
('Innodb_buffer_pool_pages_free','GAUGE'),
('Innodb_buffer_pool_pages_dirty','GAUGE'),
('Innodb_buffer_pool_pages_data','GAUGE'),
('Bytes_received','COUNTER'),
('Bytes_sent','COUNTER'),
('Innodb_rows_deleted','COUNTER'),
('Innodb_rows_inserted','COUNTER'),
('Innodb_rows_read','COUNTER'),
('Innodb_rows_updated','COUNTER'),
('Innodb_os_log_fsyncs','COUNTER'),
('Innodb_os_log_written','COUNTER'),
('Created_tmp_disk_tables','COUNTER'),
('Created_tmp_tables','COUNTER'),
('Connections','COUNTER'),
('Innodb_log_waits','COUNTER'),
('Slow_queries','COUNTER'),
('Binlog_cache_disk_use','COUNTER')
]
for _key,falcon_type in monitor_keys:
if _key == 'MySQL_QPS':
_value = int(stat_info.get('Com_select',0)) + int(stat_info.get('Qcache_hits',0))
elif _key == 'MySQL_TPS':
_value = int(stat_info.get('Com_insert',0)) + int(stat_info.get('Com_update',0)) + int(stat_info.get('Com_delete',0)) + int(stat_info.get('Com_replace',0))
elif _key == 'Innodb_buffer_read_hit_ratio':
try:
_value = round((int(stat_info.get('Innodb_buffer_pool_read_requests',0)) - int(stat_info.get('Innodb_buffer_pool_reads',0)))/int(stat_info.get('Innodb_buffer_pool_read_requests',0)) * 100,3)
except ZeroDivisionError:
_value = 0
elif _key == 'ReadWrite_ratio':
try:
_value = round((int(stat_info.get('Com_select',0)) + int(stat_info.get('Qcache_hits',0)))/(int(stat_info.get('Com_insert',0)) + int(stat_info.get('Com_update',0)) + int(stat_info.get('Com_delete',0)) + int(stat_info.get('Com_replace',0))),2)
except ZeroDivisionError:
_value = 0
else:
_value = int(stat_info.get(_key,0))
falcon_format = {
'Metric': '%s' % (_key),
'Endpoint': endpoint,
'Timestamp': timestamp,
'Step': step,
'Value': _value,
'CounterType': falcon_type,
'TAGS': tags
}
mysql_stat_list.append(falcon_format)
#_key : History list length
for _key,_value in engine_info.items():
_key = "Undo_Log_Length"
falcon_format = {
'Metric': '%s' % (_key),
'Endpoint': endpoint,
'Timestamp': timestamp,
'Step': step,
'Value': int(_value),
'CounterType': "GAUGE",
'TAGS': tags
}
mysql_stat_list.append(falcon_format)
print json.dumps(mysql_stat_list,sort_keys=True,indent=4)
requests.post(open_falcon_api, data=json.dumps(mysql_stat_list))
指标说明：收集指标里的COUNTER表示每秒执行次数，GAUGE表示直接输出值。
| 指标 | 类型 | 说明 |
| Undo_Log_Length | GAUGE | 未清除的Undo事务数 |
| Com_select | COUNTER | select/秒=QPS |
| Com_insert | COUNTER | insert/秒 |
| Com_update | COUNTER | update/秒 |
| Com_delete | COUNTER | delete/秒 |
| Com_replace | COUNTER | replace/秒 |
| MySQL_QPS | COUNTER | QPS |
| MySQL_TPS | COUNTER | TPS |
| ReadWrite_ratio | GAUGE | 读写比例 |
| Innodb_buffer_pool_read_requests | COUNTER | innodb buffer pool 读次数/秒 |
| Innodb_buffer_pool_reads | COUNTER | Disk 读次数/秒 |
| Innodb_buffer_read_hit_ratio | GAUGE | innodb buffer pool 命中率 |
| Innodb_buffer_pool_pages_flushed | COUNTER | innodb buffer pool 刷写到磁盘的页数/秒 |
| Innodb_buffer_pool_pages_free | GAUGE | innodb buffer pool 空闲页的数量 |
| Innodb_buffer_pool_pages_dirty | GAUGE | innodb buffer pool 脏页的数量 |
| Innodb_buffer_pool_pages_data | GAUGE | innodb buffer pool 数据页的数量 |
| Bytes_received | COUNTER | 接收字节数/秒 |
| Bytes_sent | COUNTER | 发送字节数/秒 |
| Innodb_rows_deleted | COUNTER | innodb表删除的行数/秒 |
| Innodb_rows_inserted | COUNTER | innodb表插入的行数/秒 |
| Innodb_rows_read | COUNTER | innodb表读取的行数/秒 |
| Innodb_rows_updated | COUNTER | innodb表更新的行数/秒 |
| Innodb_os_log_fsyncs | COUNTER | Redo Log fsync次数/秒 |
| Innodb_os_log_written | COUNTER | Redo Log 写入的字节数/秒 |
| Created_tmp_disk_tables | COUNTER | 创建磁盘临时表的数量/秒 |
| Created_tmp_tables | COUNTER | 创建内存临时表的数量/秒 |
| Connections | COUNTER | 连接数/秒 |
| Innodb_log_waits | COUNTER | innodb log buffer不足等待的数量/秒 |
| Slow_queries | COUNTER | 慢查询数/秒 |
| Binlog_cache_disk_use | COUNTER | Binlog Cache不足的数量/秒 |
使用说明：读取配置到都数据库列表执行，配置文件格式如下(mysqldb_list.txt)：
IP,Port,User,Password,endpoint
192.168.2.21,3306,root,123,mysql-21:3306 192.168.2.88,3306,root,123,mysql-88:3306
最后执行：
python mysql_monitor.py mysqldb_list.txt
2) Redis 收集信息脚本(redis_monitor.py)
#!/bin/env python
#-*- coding:utf-8 -*-
import json
import time
import re
import redis
import requests
import fileinput
import datetime
class RedisMonitorInfo():
def __init__(self,host,port,password):
self.host = host
self.port = port
self.password = password
def stat_info(self):
try:
r = redis.Redis(host=self.host, port=self.port, password=self.password)
stat_info = r.info()
return stat_info
except Exception, e:
print (datetime.datetime.now()).strftime("%Y-%m-%d %H:%M:%S")
print e
return dict()
def cmdstat_info(self):
try:
r = redis.Redis(host=self.host, port=self.port, password=self.password)
cmdstat_info = r.info('Commandstats')
return cmdstat_info
except Exception, e:
print (datetime.datetime.now()).strftime("%Y-%m-%d %H:%M:%S")
print e
return dict()
if __name__ == '__main__':
open_falcon_api = 'http://192.168.200.86:1988/v1/push'
db_list= []
for line in fileinput.input():
db_list.append(line.strip())
for db_info in db_list:
# host,port,password,endpoint,metric = db_info.split(',')
host,port,password,endpoint = db_info.split(',')
timestamp = int(time.time())
step = 60
falcon_type = 'COUNTER'
# tags = "port=%s" %port
tags = ""
conn = RedisMonitorInfo(host,port,password)
#查看各个命令每秒执行次数
redis_cmdstat_dict = {}
redis_cmdstat_list = []
cmdstat_info = conn.cmdstat_info()
for cmdkey in cmdstat_info:
redis_cmdstat_dict[cmdkey] = cmdstat_info[cmdkey]['calls']
for _key,_value in redis_cmdstat_dict.items():
falcon_format = {
'Metric': '%s' % (_key),
'Endpoint': endpoint,
'Timestamp': timestamp,
'Step': step,
'Value': int(_value),
'CounterType': falcon_type,
'TAGS': tags
}
redis_cmdstat_list.append(falcon_format)
#查看Redis各种状态,根据需要增删监控项,str的值需要转换成int
redis_stat_list = []
monitor_keys = [
('connected_clients','GAUGE'),
('blocked_clients','GAUGE'),
('used_memory','GAUGE'),
('used_memory_rss','GAUGE'),
('mem_fragmentation_ratio','GAUGE'),
('total_commands_processed','COUNTER'),
('rejected_connections','COUNTER'),
('expired_keys','COUNTER'),
('evicted_keys','COUNTER'),
('keyspace_hits','COUNTER'),
('keyspace_misses','COUNTER'),
('keyspace_hit_ratio','GAUGE'),
('keys_num','GAUGE'),
]
stat_info = conn.stat_info()
for _key,falcon_type in monitor_keys:
#计算命中率
if _key == 'keyspace_hit_ratio':
try:
_value = round(float(stat_info.get('keyspace_hits',0))/(int(stat_info.get('keyspace_hits',0)) + int(stat_info.get('keyspace_misses',0))),4)*100
except ZeroDivisionError:
_value = 0
#碎片率是浮点数
elif _key == 'mem_fragmentation_ratio':
_value = float(stat_info.get(_key,0))
#拿到key的数量
elif _key == 'keys_num':
_value = 0
for i in range(16):
_key = 'db'+str(i)
_num = stat_info.get(_key)
if _num:
_value += int(_num.get('keys'))
_key = 'keys_num'
#其他的都采集成counter，int
else:
try:
_value = int(stat_info[_key])
except:
continue
falcon_format = {
'Metric': '%s' % (_key),
'Endpoint': endpoint,
'Timestamp': timestamp,
'Step': step,
'Value': _value,
'CounterType': falcon_type,
'TAGS': tags
}
redis_stat_list.append(falcon_format)
load_data = redis_stat_list+redis_cmdstat_list
print json.dumps(load_data,sort_keys=True,indent=4)
requests.post(open_falcon_api, data=json.dumps(load_data))
指标说明：收集指标里的COUNTER表示每秒执行次数，GAUGE表示直接输出值。
| 指标 | 类型 | 说明 |
| connected_clients | GAUGE | 连接的客户端个数 |
| blocked_clients | GAUGE | 被阻塞客户端的数量 |
| used_memory | GAUGE | Redis分配的内存的总量 |
| used_memory_rss | GAUGE | OS分配的内存的总量 |
| mem_fragmentation_ratio | GAUGE | 内存碎片率，used_memory_rss/used_memory |
| total_commands_processed | COUNTER | 每秒执行的命令数，比较准确的QPS |
| rejected_connections | COUNTER | 被拒绝的连接数/秒 |
| expired_keys | COUNTER | 过期KEY的数量/秒 |
| evicted_keys | COUNTER | 被驱逐KEY的数量/秒 |
| keyspace_hits | COUNTER | 命中KEY的数量/秒 |
| keyspace_misses | COUNTER | 未命中KEY的数量/秒 |
| keyspace_hit_ratio | GAUGE | KEY的命中率 |
| keys_num | GAUGE | KEY的数量 |
| cmd_* | COUNTER | 各种名字都执行次数/秒 |
使用说明：读取配置到都数据库列表执行，配置文件格式如下(redisdb_list.txt)：
IP,Port,Password,endpoint
192.168.1.56,7021,zhoujy,redis-56:7021 192.168.1.55,7021,zhoujy,redis-55:7021
最后执行：
python redis_monitor.py redisdb_list.txt
3) MongoDB 收集信息脚本(mongodb_monitor.py)
...后续添加
4）其他相关的监控（需要装上agent），比如下面的指标：
| 告警项 | 触发条件 | 备注 |
|---|---|---|
| load.1min | all(#3)>10 | Redis服务器过载，处理能力下降 |
| cpu.idle | all(#3)<10 | CPU idle过低，处理能力下降 |
| df.bytes.free.percent | all(#3)<20 | 磁盘可用空间百分比低于20%，影响从库RDB和AOF持久化 |
| mem.memfree.percent | all(#3)<15 | 内存剩余低于15%，Redis有OOM killer和使用swap的风险 |
| mem.swapfree.percent | all(#3)<80 | 使用20% swap,Redis性能下降或OOM风险 |
| net.if.out.bytes | all(#3)>94371840 | 网络出口流量超90MB,影响Redis响应 |
| net.if.in.bytes | all(#3)>94371840 | 网络入口流量超90MB,影响Redis响应 |
| disk.io.util | all(#3)>90 | 磁盘IO可能存负载，影响从库持久化和阻塞写 |
相关文档：
https://github.com/iambocai/falcon-monit-scripts(redis monitor)
https://github.com/ZhuoRoger/redismon(redis monitor)
https://www.cnblogs.com/zhoujinyi/p/6645104.html
