Runtime options with Memory, CPUs, and GPUs

Estimated reading time: 16 minutes

By default, a container has no resource constraints and can use as much of a given resource as the host’s kernel scheduler allows. Docker provides ways to control how much memory, or CPU a container can use, setting runtime configuration flags of the docker run command. This section provides details on when you should set such limits and the possible implications of setting them.

Many of these features require your kernel to support Linux capabilities. To check for support, you can use the docker info command. If a capability is disabled in your kernel, you may see a warning at the end of the output like the following:

WARNING: No swap limit support

Consult your operating system’s documentation for enabling them. Learn more.

Memory

Understand the risks of running out of memory

It is important not to allow a running container to consume too much of the host machine’s memory. On Linux hosts, if the kernel detects that there is not enough memory to perform important system functions, it throws an OOME, or Out Of Memory Exception, and starts killing processes to free up memory. Any process is subject to killing, including Docker and other important applications. This can effectively bring the entire system down if the wrong process is killed.

Docker attempts to mitigate these risks by adjusting the OOM priority on the Docker daemon so that it is less likely to be killed than other processes on the system. The OOM priority on containers is not adjusted. This makes it more likely for an individual container to be killed than for the Docker daemon or other system processes to be killed. You should not try to circumvent these safeguards by manually setting --oom-score-adj to an extreme negative number on the daemon or a container, or by setting --oom-kill-disable on a container.

For more information about the Linux kernel’s OOM management, see Out of Memory Management.

You can mitigate the risk of system instability due to OOME by:

• Perform tests to understand the memory requirements of your application before placing it into production.
• Ensure that your application runs only on hosts with adequate resources.
• Limit the amount of memory your container can use, as described below.
• Be mindful when configuring swap on your Docker hosts. Swap is slower and less performant than memory but can provide a buffer against running out of system memory.
• Consider converting your container to a service, and using service-level constraints and node labels to ensure that the application runs only on hosts with enough memory

Limit a container’s access to memory

Docker can enforce hard memory limits, which allow the container to use no more than a given amount of user or system memory, or soft limits, which allow the container to use as much memory as it needs unless certain conditions are met, such as when the kernel detects low memory or contention on the host machine. Some of these options have different effects when used alone or when more than one option is set.

Most of these options take a positive integer, followed by a suffix of b, k, m, g, to indicate bytes, kilobytes, megabytes, or gigabytes.

For more information about cgroups and memory in general, see the documentation for Memory Resource Controller.

--memory-swap details

--memory-swap is a modifier flag that only has meaning if --memory is also set. Using swap allows the container to write excess memory requirements to disk when the container has exhausted all the RAM that is available to it. There is a performance penalty for applications that swap memory to disk often.

Its setting can have complicated effects:

• If --memory-swap is set to a positive integer, then both --memory and --memory-swap must be set. --memory-swap represents the total amount of memory and swap that can be used, and --memory controls the amount used by non-swap memory. So if --memory="300m" and --memory-swap="1g", the container can use 300m of memory and 700m (1g - 300m) swap.

• If --memory-swap is set to 0, the setting is ignored, and the value is treated as unset.

• If --memory-swap is set to the same value as --memory, and --memory is set to a positive integer, the container does not have access to swap. See Prevent a container from using swap.

• If --memory-swap is unset, and --memory is set, the container can use as much swap as the --memory setting, if the host container has swap memory configured. For instance, if --memory="300m" and --memory-swap is not set, the container can use 600m in total of memory and swap.

• If --memory-swap is explicitly set to -1, the container is allowed to use unlimited swap, up to the amount available on the host system.

• Inside the container, tools like free report the host’s available swap, not what’s available inside the container. Don’t rely on the output of free or similar tools to determine whether swap is present.

Prevent a container from using swap

If --memory and --memory-swap are set to the same value, this prevents containers from using any swap. This is because --memory-swap is the amount of combined memory and swap that can be used, while --memory is only the amount of physical memory that can be used.

--memory-swappiness details

• A value of 0 turns off anonymous page swapping.
• A value of 100 sets all anonymous pages as swappable.
• By default, if you do not set --memory-swappiness, the value is inherited from the host machine.

--kernel-memory details

Kernel memory limits are expressed in terms of the overall memory allocated to a container. Consider the following scenarios:

• Unlimited memory, unlimited kernel memory: This is the default behavior.
• Unlimited memory, limited kernel memory: This is appropriate when the amount of memory needed by all cgroups is greater than the amount of memory that actually exists on the host machine. You can configure the kernel memory to never go over what is available on the host machine, and containers which need more memory need to wait for it.
• Limited memory, unlimited kernel memory: The overall memory is limited, but the kernel memory is not.
• Limited memory, limited kernel memory: Limiting both user and kernel memory can be useful for debugging memory-related problems. If a container is using an unexpected amount of either type of memory, it runs out of memory without affecting other containers or the host machine. Within this setting, if the kernel memory limit is lower than the user memory limit, running out of kernel memory causes the container to experience an OOM error. If the kernel memory limit is higher than the user memory limit, the kernel limit does not cause the container to experience an OOM.

When you turn on any kernel memory limits, the host machine tracks “high water mark” statistics on a per-process basis, so you can track which processes (in this case, containers) are using excess memory. This can be seen per process by viewing /proc/<PID>/status on the host machine.

CPU

By default, each container’s access to the host machine’s CPU cycles is unlimited. You can set various constraints to limit a given container’s access to the host machine’s CPU cycles. Most users use and configure the default CFS scheduler. You can also configure the realtime scheduler.

Configure the default CFS scheduler

The CFS is the Linux kernel CPU scheduler for normal Linux processes. Several runtime flags allow you to configure the amount of access to CPU resources your container has. When you use these settings, Docker modifies the settings for the container’s cgroup on the host machine.

If you have 1 CPU, each of the following commands guarantees the container at most 50% of the CPU every second.

docker run -it --cpus=".5" ubuntu /bin/bash

Which is the equivalent to manually specifying --cpu-period and --cpu-quota;

$ docker run -it --cpu-period=100000 --cpu-quota=50000 ubuntu /bin/bash

Configure the realtime scheduler

You can configure your container to use the realtime scheduler, for tasks which cannot use the CFS scheduler. You need to make sure the host machine’s kernel is configured correctly before you can configure the Docker daemon or configure individual containers.

Warning

CPU scheduling and prioritization are advanced kernel-level features. Most users do not need to change these values from their defaults. Setting these values incorrectly can cause your host system to become unstable or unusable.

Configure the host machine’s kernel

Verify that CONFIG_RT_GROUP_SCHED is enabled in the Linux kernel by running zcat /proc/config.gz | grep CONFIG_RT_GROUP_SCHED or by checking for the existence of the file /sys/fs/cgroup/cpu.rt_runtime_us. For guidance on configuring the kernel realtime scheduler, consult the documentation for your operating system.

Configure the Docker daemon

To run containers using the realtime scheduler, run the Docker daemon with the --cpu-rt-runtime flag set to the maximum number of microseconds reserved for realtime tasks per runtime period. For instance, with the default period of 1000000 microseconds (1 second), setting --cpu-rt-runtime=950000 ensures that containers using the realtime scheduler can run for 950000 microseconds for every 1000000-microsecond period, leaving at least 50000 microseconds available for non-realtime tasks. To make this configuration permanent on systems which use systemd, see Control and configure Docker with systemd.

Configure individual containers

You can pass several flags to control a container’s CPU priority when you start the container using docker run. Consult your operating system’s documentation or the ulimit command for information on appropriate values.

The following example command sets each of these three flags on a debian:jessie container.

$ docker run -it \
    --cpu-rt-runtime=950000 \
    --ulimit rtprio=99 \
    --cap-add=sys_nice \
    debian:jessie

If the kernel or Docker daemon is not configured correctly, an error occurs.

GPU

Access an NVIDIA GPU

Prerequisites

Visit the official NVIDIA drivers page to download and install the proper drivers. Reboot your system once you have done so.

Verify that your GPU is running and accessible.

Install nvidia-container-runtime

Follow the instructions at (https://nvidia.github.io/nvidia-container-runtime/) and then run this command:

$ apt-get install nvidia-container-runtime

Ensure the nvidia-container-runtime-hook is accessible from $PATH.

$ which nvidia-container-runtime-hook

Restart the Docker daemon.

Expose GPUs for use

Include the --gpus flag when you start a container to access GPU resources. Specify how many GPUs to use. For example:

$ docker run -it --rm --gpus all ubuntu nvidia-smi

Exposes all available GPUs and returns a result akin to the following:

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.130            	Driver Version: 384.130               	|
|-------------------------------+----------------------+----------------------+
| GPU  Name 	   Persistence-M| Bus-Id    	Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GRID K520       	Off  | 00000000:00:03.0 Off |                  N/A |
| N/A   36C	P0    39W / 125W |  	0MiB /  4036MiB |      0%  	Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU   	PID   Type   Process name                         	Usage  	|
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Use the device option to specify GPUs. For example:

$ docker run -it --rm --gpus device=GPU-3a23c669-1f69-c64e-cf85-44e9b07e7a2a ubuntu nvidia-smi

Exposes that specific GPU.

$ docker run -it --rm --gpus '"device=0,2"' ubuntu nvidia-smi

Exposes the first and third GPUs.

Note

NVIDIA GPUs can only be accessed by systems running a single engine.

Set NVIDIA capabilities

You can set capabilities manually. For example, on Ubuntu you can run the following:

$ docker run --gpus 'all,capabilities=utility' --rm ubuntu nvidia-smi

This enables the utility driver capability which adds the nvidia-smi tool to the container.

Capabilities as well as other configurations can be set in images via environment variables. More information on valid variables can be found at the nvidia-container-runtime GitHub page. These variables can be set in a Dockerfile.

You can also utitize CUDA images which sets these variables automatically. See the CUDA images GitHub page for more information.