[PYTHON] [blackbird-rds] Monitoring AWS RDS
blackbird RDS plugin
The source code is here
This plugin nicely gets the RDS Metric from CloudWatch. To be honest, I can't deny the feeling of wearing MySQL plugin, but you can get the contents like Summary.
You Can Get Following Metrics
Then, from what you can get with this plugin immediately.
| CloudWatch Metric | Statistics | Detail |
|---|---|---|
| BinLog Disk Usage | Average | Disk space used by binlog(bytes) |
| CPU Utilization | Average | CPU usage(%) |
| Database Connections | Average | Number of connections to the current DB |
| Disk Queue Depth | Average | Accumulated in Queue(Unprocessed)I/Number of O processes |
| FreeStorageSpace | Average | Free disk space(bytes) |
| Feeable Memory | Average | Available memory capacity(bytes) //I use it quite a bit with InnodbBufferPool, so it doesn't matter if it's small. |
| NetworkReceiveThroughput | Average | Inbound network traffic on RDS server |
| NetworkTransmitThroughput | Average | Outgoing network traffic on RDS server |
| ReplicaLag | Average | Slave's replica delay, so-called seconds behind master(seconds) |
| SwapUsage | Average | Amount of swapping from memory(bytes) |
| ReadIOPS | Average | Read IOPS(If it is Provisioned, it will stop at exactly that number) |
| WriteIOPS | Average | Write IOPS(If it is Provisioned, it will stop at exactly that number) |
| ReadLatency | Average | Read Latency(The unit is ms) |
| WriteLatency | Average | Write Latency(The unit is ms) |
| ReadThroughput | Average | Throughput of Read, how many bytes were read per second |
| WriteThroughput | Average | Throughput of Write, how many bytes were written per second |
I thought it was written, but it really feels like a summary of SHOW GLOBAL STATUS or SHOW ENGINE INNODB STATUS. It's a pity that I can't get QueryPerSecond or Innodb Cache Hit Rate (that's a MySQL plugin).
Zabbix Template
Items
Since there was something strict in monitoring with the above Metric alone, I tried my best on the calculation item and plugin side so that the following values can also be obtained.
| Item Name Outline | Detail |
|---|---|
| Total Memory Size | Calculate Total Memory Size from instance type and Memory mapping |
| Used Memory Size | Total Memory Size - Freeable MemoryCalculated by |
| Percent of Memory Available | Free memory space% |
| Total Disk Size | Calculated from the mapping between instance type and Disk capacity |
| Used Disk Size | Total Disk Size - Free Storage SpaceCalculated by |
| Percent of Disk Available | Free disk space% |
// Honestly, Amazon, I wanted you to have an API that can get the mapping between the instance type and CPU, Memory, Disk ...
Triggers
- Is CPU Utilization exceeding a certain number?
- Is the disk usage rate exceeding a certain number?
- Whether the Disk is SwapIn
So far, this is the only one, but honestly, considering the InnodbBufferPool, it doesn't make sense to make it an alert, isn't it just a delay of Slave? I want you to tell me!
Graphs
Disk Usage
Memory Usage
Database Connections
Omitted because it is an ordinary bar graph
NetworkTraffic
CPU Utilization
Ry for ordinary bar charts
Disk I/O Latency
Disk I/O Queue Depth
Ry for ordinary bar charts
Disk I/O Throughput
Because it is very similar to the graph of Disk I / O Queue Depth, ry
IOPS
Because it is very similar to the graph of Disk I / O Queue Depth, ry
Replica Lag (Seconds Behind Master)
Ry for ordinary bar charts
It looks like this as a template of Zabbix. If you line up this on the Screen, you can see at a glance which RDS is busy (by the way, I monitor it that way).
How to Install
Case of pip
pip install blackbird-rds
that's all.
Case of yum
Please refer to here for the RPM repository of blackbird.
yum install blackbird-rds --enablerepo=blackbird
Configuration Your blackbird
#Since the section name is used internally for the Thread name, anything is okay, but it is safer not to wear it.
[ANYTHING_OK]
#The acquisition interval and minimum value are 60 seconds, and even if set to 1 second, it will be 60 seconds.
interval = 60
#AWS region
region_name = ap-northeast-1
#AWS credential
aws_access_key_id = YOUR_AWS_ACCESS_KEY_ID
aws_secret_access_key = YOUR_AWS_SECRET_ACESS_KEY
#This is the DB name given when creating the RDS instance.
db_instance_identifier = YOUR_RDS_INSTANCE_NAME
#Host name on zabbix server. Don't forget to make it first.
hostname = prod-dbidm01
module = rds
#This value is empty by default. However, if you want to save money by reducing the number of API calls in CloudWatch, write them separated by commas as follows. Then, the specified Metric will not be acquired.
ignore_metrics = ReplicaLag,BinLogDiskUsage
Recommended Posts