How to fix Error response from daemon: Container is not running in Docker

First, verify the container's current state to understand why it's not running:

docker ps -a

Look for your container in the output. The STATUS column will show either:
- "Up X minutes" (running - this error shouldn't occur)
- "Exited (0)" (completed successfully)
- "Exited (1)" or other non-zero code (crashed or error)
- "Created" (never started)

Note the container ID or name for the next steps.

If the container has exited, restart it first:

docker start <container_name_or_id>

Then immediately run your exec command:

docker exec -it <container_name_or_id> /bin/bash

Note: If the container exits again immediately after starting, proceed to the next steps to diagnose why.

If the container keeps exiting, check its logs to understand why:

docker logs <container_name_or_id>

Common issues you might find:
- Application crashes or exceptions
- Missing environment variables or configuration
- File not found errors
- Permission denied errors

The exit code also provides hints:
- Exit code 0: Process completed successfully (not an error, just finished)
- Exit code 1: General error
- Exit code 137: Killed by SIGKILL (often OOM killer)
- Exit code 139: Segmentation fault

For images designed around shells (like ubuntu, alpine, debian), you need to allocate a terminal to keep them running:

# Run with interactive terminal
docker run -it --name my_container <image_name> /bin/bash

# Or run detached with a pseudo-TTY to keep alive
docker run -d -t --name my_container <image_name>

# Now you can exec into it
docker exec -it my_container /bin/bash

The -t flag allocates a pseudo-TTY, and -i keeps STDIN open. Together they keep shell-based containers running.

If you need a container to stay running for debugging or testing, override the entrypoint with a long-running process:

# Run with tail -f to keep container alive
docker run -d --name my_container <image_name> tail -f /dev/null

# Or use sleep infinity
docker run -d --name my_container <image_name> sleep infinity

# Now exec works
docker exec -it my_container /bin/bash

This is useful when the original CMD/ENTRYPOINT exits too quickly for you to debug.

If you only need to inspect files in a stopped container (not run commands), you can commit and create a new container:

# Commit the stopped container to a new image
docker commit <stopped_container_id> debug-image

# Run a new container from that image with a shell
docker run -it debug-image /bin/bash

Alternatively, copy files directly from the stopped container:

docker cp <stopped_container_id>:/path/to/file ./local_destination