Container metrics collection in Docker - Step-by-Step Execution

Process Flow - Container metrics collection

Start Container

↓

Enable Metrics Collection

↓

Collect Metrics Data

↓

Store Metrics in Database

↓

Query Metrics for Analysis

↓

Use Metrics for Monitoring/Alerting

This flow shows how container metrics are collected, stored, and used for monitoring.

Execution Sample

Docker

docker stats --no-stream --format "{{.Container}} {{.CPUPerc}} {{.MemUsage}}"
INSERT INTO container_metrics (container_id, cpu_percent, memory_usage) VALUES (?, ?, ?);
SELECT * FROM container_metrics WHERE container_id = ?;

This code collects live container metrics, stores them in a database, and queries them.

Process Table

Step	Action	Input/Command	Output/Result	Notes
1	Start container	docker run -d nginx	Container ID: abc123	Container abc123 is running
2	Collect metrics	docker stats --no-stream --format "{{.Container}} {{.CPUPerc}} {{.MemUsage}}"	abc123 0.05% 10MiB / 500MiB	Metrics snapshot taken
3	Insert metrics	INSERT INTO container_metrics VALUES ('abc123', 0.05, '10MiB / 500MiB')	1 row inserted	Metrics stored in DB
4	Query metrics	SELECT * FROM container_metrics WHERE container_id = 'abc123'	abc123 | 0.05 | 10MiB / 500MiB	Retrieved stored metrics
5	End	-	-	Process complete

💡 Metrics collected once; process ends after storing and querying metrics.

Status Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	Final
container_id	null	abc123	abc123	abc123	abc123	abc123
cpu_percent	null	null	0.05	0.05	0.05	0.05
memory_usage	null	null	10MiB / 500MiB	10MiB / 500MiB	10MiB / 500MiB	10MiB / 500MiB

Key Moments - 3 Insights

Why do we use --no-stream with docker stats?

Using --no-stream makes docker stats output a single snapshot instead of continuous updates, as shown in step 2 of the execution_table.

Why do we store metrics in a database after collecting them?

Storing metrics allows querying and historical analysis later, demonstrated in steps 3 and 4 where metrics are inserted and then retrieved.

What does the memory usage value '10MiB / 500MiB' mean?

It shows current memory used (10MiB) out of total available (500MiB), as seen in step 2 output.

Visual Quiz - 3 Questions

Look at the execution_table, what is the container_id after step 1?

A null

B abc123

C nginx

D 0.05

💡 Hint

Check the 'Output/Result' column in row for step 1.

At which step is the metrics data inserted into the database?

A Step 3

B Step 2

C Step 4

D Step 5

💡 Hint

Look for the 'Insert metrics' action in the execution_table.

If we remove --no-stream from docker stats, what changes in the execution?

A Container would stop running

B Metrics would not be collected at all

C Metrics would be collected continuously, not just once

D Database insertion would fail

💡 Hint

Refer to the key_moments explanation about --no-stream option.

Concept Snapshot

Container metrics collection:
- Start container
- Use 'docker stats --no-stream' for one-time metrics
- Parse output (container ID, CPU %, memory)
- Store metrics in database
- Query metrics for monitoring or analysis

Full Transcript

This visual execution trace shows how container metrics are collected using docker commands, stored in a database, and queried for analysis. First, a container is started and identified by its container ID. Then, the 'docker stats' command with the --no-stream option collects a single snapshot of CPU and memory usage. These metrics are inserted into a database table for storage. Finally, the stored metrics can be queried to retrieve the data for monitoring or alerting purposes. Key points include using --no-stream to avoid continuous streaming and understanding the memory usage format. This process helps monitor container performance effectively.

Practice

1. What does the docker stats command primarily show?

easy

A. Logs of a specific container

B. List of all Docker images on the system

C. Live resource usage metrics of running containers

D. Docker daemon configuration settings

Solution

1. Step 1: Understand the purpose of docker stats

   This command is designed to display live resource usage statistics such as CPU, memory, and network for running containers.

2. Step 2: Differentiate from other Docker commands

   Other commands like docker images list images, docker logs show logs, and docker info shows daemon info, so they do not match the description.

3. Final Answer:

   Live resource usage metrics of running containers -> Option C

4. Quick Check:

   docker stats = live container metrics [OK]

Hint: Remember: stats = live resource usage [OK]

Common Mistakes:

• Confusing stats with logs
• Thinking stats shows images
• Assuming stats shows config

2. Which of the following is the correct syntax to get a single snapshot of container metrics without continuous updates?

easy

A. docker stats --no-stream

B. docker stats --single

C. docker stats --once

D. docker stats --snapshot

Solution

1. Step 1: Identify the flag for single snapshot

   The --no-stream flag tells docker stats to show metrics once and exit instead of streaming live updates.

2. Step 2: Verify other options

   Flags like --single, --once, and --snapshot are not valid Docker flags for this command.

3. Final Answer:

   docker stats --no-stream -> Option A

4. Quick Check:

   Single snapshot flag = --no-stream [OK]

Hint: Use --no-stream for one-time metrics [OK]

Common Mistakes:

• Using invalid flags like --single
• Omitting the flag and expecting single output
• Confusing with other Docker commands

3. What will be the output of the command docker stats --no-stream my_container if my_container is running?

medium

A. A continuous live stream of resource usage for all containers

B. Logs of my_container

C. An error saying container not found

D. A single snapshot of resource usage metrics for my_container

Solution

1. Step 1: Understand the command components

   docker stats --no-stream requests a single snapshot, and specifying my_container limits output to that container only.

2. Step 2: Predict the output

   Since my_container is running, the command outputs one snapshot of its resource usage metrics, not a stream or logs.

3. Final Answer:

   A single snapshot of resource usage metrics for my_container -> Option D

4. Quick Check:

   --no-stream + container name = single snapshot for that container [OK]

Hint: Add container name for focused single snapshot [OK]

Common Mistakes:

• Expecting continuous stream despite --no-stream
• Confusing stats output with logs
• Assuming error if container is running

4. You run docker stats --no-stream my_container but get an error: Error: No such container: my_container. What is the most likely cause?

medium

A. You need to add --all flag to see stopped containers

B. The container my_container is not running or does not exist

C. Docker daemon is not running

D. The --no-stream flag is invalid

Solution

1. Step 1: Analyze the error message

   The error clearly states the container my_container cannot be found, meaning it is either stopped or never created.

2. Step 2: Check other options

   The --no-stream flag is valid, Docker daemon errors would be different, and docker stats does not support --all to show stopped containers.

3. Final Answer:

   The container my_container is not running or does not exist -> Option B

4. Quick Check:

   Error 'No such container' means container missing [OK]

Hint: Check container existence before stats command [OK]

Common Mistakes:

• Assuming flag is wrong
• Thinking Docker daemon is down
• Trying unsupported flags like --all

5. You want to monitor CPU and memory usage of two specific containers named webapp and db with a single snapshot. Which command should you run?

hard

A. docker stats --no-stream webapp db

B. docker stats webapp db --no-stream

C. docker stats --no-stream --filter webapp --filter db

D. docker stats --filter name=webapp,db --no-stream

Solution

1. Step 1: Understand how to specify multiple containers

   Docker stats accepts container names as arguments separated by spaces to monitor multiple containers.

2. Step 2: Use the correct flag placement

   The --no-stream flag should come before container names; docker stats --no-stream webapp db places it correctly.

3. Step 3: Evaluate other options

   docker stats webapp db --no-stream incorrectly places the flag after container names, causing Docker to interpret --no-stream as a container name and resulting in an error. Options C and D use invalid or unsupported --filter syntax for this command.

4. Final Answer:

   docker stats --no-stream webapp db -> Option A

5. Quick Check:

   Multiple containers + --no-stream = docker stats --no-stream webapp db [OK]

Hint: List containers after --no-stream for snapshot [OK]

Common Mistakes:

• Using invalid --filter syntax
• Placing flags after container names incorrectly
• Trying to monitor stopped containers