[PYTHON] [blackbird-aws-service-limits] Monitoring AWS Service Limits

blackbird AWS Service Limits plugin. This plugin gets various AWS Service Limits and registers them in zabbix. By default (overwrite with Macro), an alert will be sent when 80% of the service limit (for example, overuse of EIP or EC2 will soon reach the limit) is reached.

List where you can get the current usage

We monitor the following service limits and current usage.

• AutoScale Launch Configurations
• AutoScale Groups
• EC2 Elastic IPs
• EC2 Running Instances
• Number of ELBs
• RDS Instances
• RDS Total Storages
• Number of RDS Read Replicas per Master
• DynamoDB Read Capacity Units for Individual Table
• DynamoDB Write Capacity Units for Individual Table
• DynamoDB Read Capacity Units per Account
• DynamoDB Write Capacity Units per Account

Supplement About Read / Write Capacity Units of DynamoDB

There are two types of Service Limits in DynamoDB's Capacity Units.

• Upper limit for each table
• Account-wide limit

There is. So, the upper limit for each table is the highest Capacity Unit in the account, and the upper limit for the entire account is (Japanese is complicated and I'm sorry) ** The sum of the Capacity Units of all the tables ** Is applied. So, if you are using Kinesis, write to DynamoDB how much Kinesis Client got Records. So if you're writing or reading a lot on Kinesis Stream, be careful.

List of things with API

Actually, there may or may not be an API that can get the current upper limit setting (almost none w)

• AutoScale Launch Configurations
• AutoScale Groups
• EC2 Elastic IPs
• EC2 Running Instances

Here is a matrix of these.

Zabbix Template

Items

As for the value, you can get the same item as described in List where you can get the current usage amount.

Macros

Unfortunately, if there is no current upper limit acquisition API, it is specified in Macro ... (However, since the initial value is embedded in Macro, it will be easier to understand if you refer to that)

If you apply for the upper limit relaxation, update Macro!

about {$TRG_AWS_SVC_THR_XXXXXX_AVG}

Normally, when the upper limit is 80%, alert notification is sent with Average Severity, but no, only EC2 Running Interfaces wants notification at 60%, or conversely, since Public IP is used, EIP notification can be 90%. If you like, please overwrite this Macro. For example, overwrite it with something like {$ TRG_AWS_SVC_THR_EC2_INS_AVG} = 60 or {$ TRG_AWS_SVC_THR_EC2_EIP_AVG} = 90. By the way, the team I belong to receives only 60% notification of Read / Write Capacity Unit of DynamoDB (because I use Kinesis).

Graphs

The actual value is displayed with each upper limit as the vertex of the Y axis of the graph. I usually use this as a screen to monitor.

Notes

I'm doing Auto Scaling but it doesn't scale! I feel that it is about the upper limit. Also, EIP suddenly hits the upper limit without being aware of it. .. ..

You won't be able to see when the ELB hits ...