Bash script that reports custom metric data about PostgreSQL(RDS) performance to Amazon CloudWatch
This script is intended for use with Amazon EC2 instances running Linux operating systems for monitoring Amazon RDS for PostgreSQL instances.
The scripts have been tested on the following Amazon Machine Images (AMIs) for both 32-bit and 64-bit versions:
- Amazon Linux 2014.9
- Red Hat Enterprise Linux 6.4
- Ubuntu Server 13.10
The scripts have been tested on the following RDS engines:
- PostgreSQL 9.3.5
The script is written in simple Bash, so you can use it very easily.
This script can report the following metrics:
- status
- cache hit
- session
- transaction
- lock
- tupple read
- tupple write
- block read
- buffer write
This script requires AWS Command Line Interface.
Instances that you launch using the Amazon Linux AMI already include the CLI tools.
All you have to do is to install git and postgresql-client.
yum install git postgresql
Install git and postgresql-client.
yum install git postgresql
Install git, unzip, jre and postgresql-client.
apt-get install git unzip default-jre postgresql-client
You can install AWS Command Line Interface with following steps:
pip install awscli
Download "aws-mon-pgsql.sh" or clone repository:
git clone https://github.com/moomindani/aws-mon-pgsql.git
cd aws-mon-pgsql
Make ".pgpass" in order to connect postgres without password:
echo "postgres.xxx.ap-northeast-1.rds.amazonaws.com:5432:postgres:postgres:password" > ~/.pgpass
chmod 600 ~/.pgpass
Use the script:
./aws-mon-pgsql.sh --help
This script collects status, cache hit, session, transaction, lock, tupple read, tupple write, block read, buffer write data on the target RDS. It then makes a remote call to Amazon CloudWatch to report the collected data as custom metrics.
Name | Description |
---|---|
--help | Displays usage information. |
--version | Displays the version number. |
--verify | Checks configuration and prepares a remote call. |
--verbose | Displays details of what the script is doing. |
--debug | Displays information for debugging. |
--from-cron | Use this option when calling the script from cron. |
--profile VALUE | Use a specific profile from your credential file. |
--id | Specifies database instance identifier. |
-h | Specifies database server host. |
-p | Specifies database server port. |
-U | Specifies database user. |
-d | Specifies database name. |
--status | Reports the status of the database instance. |
--query-execution | Reports whether select query is executed successfully or not. |
--query-execution-timeout | Specifies timeout for query execution. |
--cache-hit | Reports cache hit ratio. |
--session-active | Reports the number of sessions whose status is active. |
--session-idle | Reports the number of sessions whose status is idle. |
--session-wait | Reports the number of sessions whose status is wait. |
--txn-commit | Reports the number of transactions committed. (Run at least twice with same options) |
--txn-rollback | Reports the number of transactions rollbacked. (Run at least twice with same options) |
--locks-acquired | Reports the number of locks acquired. |
--locks-wait | Reports the number of locks wait. |
--tup-inserted | Reports the number of tupples inserted. (Run at least twice with same options) |
--tup-updated | Reports the number of tupples updated. (Run at least twice with same options) |
--tup-deleted | Reports the number of tupples deleted. (Run at least twice with same options) |
--tup-returned | Reports the number of tupples returned. (Run at least twice with same options) |
--tup-fetched | Reports the number of tupples fetched. (Run at least twice with same options) |
--blks-read | Reports the number of blocks not included in shared memory and read from disk. (Needs to run at least twice with same options) |
--blks-hit | Reports the number of blocks included in shared memory. (Run at least twice with same options) |
--buffers-checkpoint | Reports the number of buffers written for checkpoint. (Run at least twice with same options) |
--buffers-clean | Reports the number of buffers written for cleaning. (Run at least twice with same options) |
--buffers-backend | Reports the number of buffers written for backend. (Run at least twice with same options) |
--all-items | Reports all items. |
The following examples assume that you have already configured awscli with "aws configure".
- Run the following command:
./aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --status-check-failed --verify --verbose
- Run the following command:
./aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --status-check-failed --verbose
- Run the following command:
./aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --session-active --session-idle --session-wait --verbose
- Run the following command:
./aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --cache-hit --verbose
- Run the following command:
./aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --tup-inserted --tup-updated --tup-deleted --tup-returned --tup-fetched --verbose
(wait a few seconds...)
./aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --tup-inserted --tup-updated --tup-deleted --tup-returned --tup-fetched --verbose
- Start editing the crontab using the following command:
crontab -e
- Add the following command to report all items to CloudWatch every five minutes:
*/5 * * * * ~/aws-mon-pgsql/aws-mon-pgsql.sh --id postgres -h postgres.xxx.ap-northeast-1.rds.amazonaws.com -p 5432 -U postgres -d postgres --all-items --from-cron