PEM on Docker is not officially supported. I recommend to use this demo instead.

This demo is built for Docker and is using the Trusted Postgres Architect (TPA) to deploy a 2-node cluster managed by EDB Failover Manager, a Barman server and a PEM server. The objective of the demo environment is to show the value of Postgres Enterprise Manager.

TPA will deploy the following components:

The EFM cluster which is created is called pemdemo.

Status of the EFM cluster can be shown using docker exec -it pg1 bash -c "/usr/edb/efm-4.7/bin/efm cluster-status pemdemo"

Run 00-provision.sh to provision the Postgres containers (pg1 and pg2), the barman container (barman) and the PEM container (pemserver). This deployment will take appx. 20 minutes to complete. After successful deployment PEM should be available on https://localhost/pem. Sometimes it takes a few minutes for the PEM container to fully stabelize. You can see that happening in your Docker Desktop Dashboard where the CPU of the container is still spiking. Wait for the CPU to stabelize before to continue.

PEM user is enterprisedb and the access password for this user can be revealed using tpaexec show-password pemdemo enterprisedb. I suggest you copy this password on your clipboard because you will need it in various places.

Important: After setting up the demo you need to disconnect from PG1 and PG2, add the EFM parameters to Propeties / Advanced.

EFM cluster name : pemdemo
EFM installation path : /usr/edb/efm-4.7/bin/

This enables you to use the streaming replication dashboard.

Another enhancement would be to set up a cron job which runs pgbench like this:

0,30 * * * * (PGPASSWORD='&I$iHuprYGOljC1CKoljC7H%7$HTmLLl' pgbench -h localhost -p 5444 -T 100 -c 10 -j 2 -U enterprisedb postgres) 2>&1 |logger -t pgbench

Pgbench is already initialized into the postgres database by the provisioning script.

Open a broweser, go to http://localhost/pem and log in using user enterprisedb and the password you got at the end of the provisioning process.

Give an overview of the UI and the dashboards.

• Select Monitoring
• Select Global Overview / pg1 / Alerts. You see one alert. Deep-dive into that alert.
• Select pg1 / Alerts. Click on the alert (Swap consumption percentage) and explain settings. Explain notification methods.
• Acknowledge the alert.
• Select Home / Alerts to show all alerts.

• Right-click on pg1 and select Management / Manage Probes
• Show some system probes (eg. Database Statistics)
• Show table pemdata.table_statistics which contains the same data.
• Right-click on pg1 and select Management / Manage Alerts.
• Select Alert Templates, Database size and press th pencil.
• Show Probe dependency and the tabe used in the SQL tab.
• Show Alert Templates, Email Templates.

• select pg1 and select Tools / Server / SQL Profiler / Create Trace
• Enter trace details
• Right-click on database postgres on pg1 and select Query Tool
• Create a table using create table t_test(id serial, name text);.
• Generate data using insert into t_test(name) select 'Test' from generate_series(1,2000000);
• Retrieve a record using select id from t_test where id=1234567;
• Select the SQL Profiler tab and select the query from the log. You can use the filter for this.
• Click on the Table icon in the plan and notice the node type and the cost of the query (10266.67)
• Open the index Advisor (graph icon in the top). Notice the differenc ein Node Type.
• Select the t_test table in the Suggested indexes pane and select Ok.
• Run the same query again and find the query in the SQL Profile pane again. Notice the Node Type and the total cost (4.45).

• Select pg1 and from the top meno, select the overall dashboard.
• Open a teminal and get the password for user dba using taexec show-password pemdemo dba.
• In the same terminal, generate traffic using pgbench -h localhost -p 5444 -i -U dba postgres and then pgbench -h localhost -p 5444 -T 100 -c 10 -j 2 -U dba postgres.
• Select Tools / Server / Performance diagostics and walk through the Wait events dertails options.

Show barman graphs

• Select the Barman server and select the dashboard.

Open database pem on the pemserver and show three schemas for PEM configuration (pem), PEM data (pemdata) and PEM hostorical data (pemhstory).

• Right-click on pem/Tables/agent and select Edit/View data / All rows.
• Right-click on agent and select Query Tool and show that you can query the agent table using select * from pem.agent.
• Do the same for agent_config
• Do the same for several tables in the pemdata schema. Specifically show cpu_usage
• Show cpu-usage in pemhistory and explain the difference between pemdata and pemhistory.

To clean up the demo environment you just have to run 99-deprovision.sh. This script will remove the docker containers and the cluster configuration.

This demo is broken on Docker Engine V25 and beyond (eg. Docker Desktop for Mac v4.27 and later). You will get the following error:

TASK [sys : Enable rc-local service] *********************************************************************************************************************************************************************************************************
fatal: [pg1]: FAILED! => {"changed": false, "cmd": "/usr/bin/systemctl", "msg": "Failed to connect to bus: No such file or directory", "rc": 1, "stderr": "Failed to connect to bus: No such file or directory\n", "stderr_lines": ["Failed to connect to bus: No such file or directory"], "stdout": "", "stdout_lines": []}

This is a Docker Engine issue and the only way i found to work around this is to run a pre-V25 Docker Engine. For Docker Desktop for Mac this is version 4.26.1.

Streaming setup

OS dashboards not working

REST API demo flow