由于微软云上的MySQL实例无法安装插件来导出数据，因此参考上一篇 通过Zabbix官方模板监控微软云Azure上多台Redis实例， 我们可以修改zabbix官方redis脚本，通过show status等命令获取到mysql的状态指标，用trap-messages模式发送给zabbix来监控微软云上的MySQL实例。
1. 运行环境：
Zabbix官方提供的redis监控模板，我们修改官方提供的python脚本来监控MySQL。
项目地址：https://github.com/blacked/zbx_redis_template
Zabbix所有官方模板：http://www.zabbix.org/wiki/Zabbix_Templates
2. 部署
trap-messages方式的说明请参考上一篇 通过Zabbix官方模板监控微软云Azure上多台Redis实例。
1> 安装Python依赖
# yum -y install python-pip # pip install argparse # pip install pymysql
2> 将zbx_redis_stats.py放到zabbix服务器上某路径下并重命名，如/etc/zabbix/script/mysql/zbx_mysql_stats.py。
3> 修改脚本，下面是我修改的脚本，监控了show status获取的信息以及数据库表占用的空间等信息，可以直接拷贝，或者添加自己需要监控的其他信息。
# !/usr/bin/python
import sys, pymysql, json, re, struct, time, socket, argparse
parser = argparse.ArgumentParser(description='Zabbix MySQL status script')
parser.add_argument('mysql_hostname', nargs='?')
parser.add_argument('metric', nargs='?')
parser.add_argument('db', default='none', nargs='?')
parser.add_argument('-p', '--port', dest='mysql_port', action='store', help='MySQL server port', default=3306, type=int)
parser.add_argument('-a', '--auth', dest='mysql_pass', action='store', help='MySQL server pass', default=None)
parser.add_argument('-u', '--user', dest='mysql_user', action='store', help='MySQL server user', default=None)
args = parser.parse_args()
zabbix_host = '127.0.0.1' # Zabbix Server IP
zabbix_port = 10051 # Zabbix Server Port
# Name of monitored server like it shows in zabbix web ui display
mysql_hostname = args.mysql_hostname if args.mysql_hostname else socket.gethostname()
class Metric(object):
def __init__(self, host, key, value, clock=None):
self.host = host
self.key = key
self.value = value
self.clock = clock
def __repr__(self):
result = None
if self.clock is None:
result = 'Metric(%r, %r, %r)' % (self.host, self.key, self.value)
else:
result = 'Metric(%r, %r, %r, %r)' % (self.host, self.key, self.value, self.clock)
return result
def value_format(value):
if value.isdigit() is True:
value = int(value)
return value
def send_to_zabbix(metrics, zabbix_host='127.0.0.1', zabbix_port=10051):
result = None
j = json.dumps
metrics_data = []
for m in metrics:
clock = m.clock or ('%d' % time.time())
metrics_data.append(
('{"host":%s,"key":%s,"value":%s,"clock":%s}') % (j(m.host), j(m.key), j(m.value), j(clock)))
json_data = ('{"request":"sender data","data":[%s]}') % (','.join(metrics_data))
data_len = struct.pack('<Q', len(json_data))
packet = 'ZBXD\x01' + data_len + json_data
# For debug:
# print(packet)
# print(':'.join(x.encode('hex') for x in packet))
try:
zabbix = socket.socket()
zabbix.connect((zabbix_host, zabbix_port))
zabbix.sendall(packet)
resp_hdr = _recv_all(zabbix, 13)
if not resp_hdr.startswith('ZBXD\x01') or len(resp_hdr) != 13:
print('Wrong zabbix response')
result = False
else:
resp_body_len = struct.unpack('<Q', resp_hdr[5:])[0]
resp_body = zabbix.recv(resp_body_len)
zabbix.close()
resp = json.loads(resp_body)
# For debug
# print(resp)
if resp.get('response') == 'success':
result = True
else:
print('Got error from Zabbix: %s' % resp)
result = False
except:
print('Error while sending data to Zabbix')
result = False
finally:
return result
def _recv_all(sock, count):
buf = ''
while len(buf) < count:
chunk = sock.recv(count - len(buf))
if not chunk:
return buf
buf += chunk
return buf
def main():
if mysql_hostname and args.metric:
client = pymysql.connect(host=mysql_hostname, user=args.mysql_user, password=args.mysql_pass)
cursor = client.cursor()
cursor.execute("show status")
data = cursor.fetchall()
dict_local = {}
for k, v in data:
dict_local[k] = v
cursor.execute("select SCHEMA_NAME from information_schema.schemata ")
db = cursor.fetchall()
dblist = []
for i in db:
dblist.append(i[0])
if args.metric:
def list_db():
if args.db in dblist:
print(dblist)
else:
print('database_detect')
def default():
if args.metric in dict_local.keys():
print(dict_local[args.metric])
{
'list_db': list_db
}.get(args.metric, default)()
else:
print('Not selected metric')
else:
client = pymysql.connect(host=mysql_hostname, user=args.mysql_user, password=args.mysql_pass)
cursor = client.cursor()
cursor.execute("show status")
data = cursor.fetchall()
a = []
for k, v in data:
k = k.lower()
v = value_format(v)
a.append(Metric(mysql_hostname, ('azuremysql[%s]' % k), v))
# cursor.execute("show processlist;")
cursor.execute("select count(*) from information_schema.processlist;")
processdata = cursor.fetchall()
# process_count = len(processdata)
process_count = processdata[0][0]
a.append(Metric(mysql_hostname, 'azuremysql[process_count]', process_count))
cursor.execute(
"select table_schema,round(sum(data_length+index_length)/1024/1024,4) from information_schema.tables group by table_schema;")
spacedata = cursor.fetchall()
totalspace = 0
for k, v in spacedata:
a.append(Metric(mysql_hostname, ('azuremysql[%s]' % k), float(v)))
totalspace += float(v)
a.append(Metric(mysql_hostname, 'azuremysql[totalspace]', totalspace))
# Send packet to zabbix
send_to_zabbix(a, zabbix_host, zabbix_port)
# for i in a:
# print(i)
cursor.close()
client.close()
if __name__ == '__main__':
main()
------------------------------------------分割线------------------------------------------
具体下载目录在 /2017年资料/10月/10日/通过修改Zabbix官方redis模板监控微软云Azure上多台MySQL实例/
------------------------------------------分割线------------------------------------------
4> 测试脚本是否可以正常连接到微软云上MySQL并获取参数：
/etc/zabbix/script/mysql/zbx_mysql_status.py -u mysqluser -a mysqlpasswd test.mysqldb.chinacloudapi.cn list_db
5> 添加crontab，定期执行脚本发送数据给zabbix，注意若密码中含有%需要用 \ 进行转义 (\%)。
若要监控多台MySQL，则添加多个crontab：
*/1 * * * * /etc/zabbix/script/mysql/zbx_mysql_status.py -u mysqluser -a mysqlpasswd test.mysqldb.chinacloudapi.cn
6> 创建MySQL监控模板，注意监控项类型应选择"zabbix采集器"，若使用的脚本，键值为"azuremysql[xxxxxxx]"，否则请使用自己脚本定义的键值。
附件是我创建的一个模板，添加了一些监控项，并不完整，可以在此基础上再添加需要的监控项。
zbx_Azure_MySQL_template.xml (下载见上面的Linux公社资源站)
7> 使用MySQL hostname创建主机，链接模板，即可实现zabbix监控。
更多Zabbix相关教程集合：