Limit docker container RAM usage
With the release of Helmut 4.8.x and the corresponding update to the Java environment, it is now possible to limit the maximum amount of RAM a container can allocate.
Setting memory limits for containers helps ensure predictable resource usage and protects the host system from excessive memory consumption.
Many modern runtimes, including Java, use garbage collection (GC) to manage memory allocation. These mechanisms generally behave more predictably when a memory boundary is defined. Without a defined limit, a container may continue allocating memory as long as the host system has available RAM.
In such situations, a single container could consume large amounts of memory and potentially impact the performance or stability of other containers running on the same host.
More information about garbage collection can be found here: https://en.wikipedia.org/wiki/Garbage_collection_(computer_science)
New Helmut4 Installation
By default, this RAM limitation setting is included in every new Helmut4 installation, whether it is deployed as a single server or a cluster system.
The RAM limits are defined as default parameters in the stack YAML file.
Updating Existing Helmut4 Installations
Existing Helmut4 installations can be updated to use the same resource allocation settings. To implement this change, users need to update the stack YAML configuration.
You can download the default stack YAML file from our repository for reference:
Single Server: https://repo.moovit24.de/install/config/docker-compose.yml
Cluster System: https://repo.moovit24.de/install/config/docker-compose-swarm.yml
Environment-Specific Adjustments
Please note that these values may need to be adjusted depending on your system requirements.
For example, systems with high I/O activity or CPU load may require additional memory for related containers such as io, co, or mongodb.
Example
Below is an example configuration showing a RAM limit for the language container:
Default RAM limits
The following values can be considered as defaults:
Environment-Specific Adjustments
The following values can be considered default recommendations.
Please note that these values may need to be adjusted depending on your system requirements. For example, systems with high I/O and CPU workloads might require increased memory for containers such as io, co, and mongodb.
traefik
250
hp
300
hw
300
fx
1500
io
2000
co
2000
hk
1500
users
1500
streams
2000
preferences
400
metadata
350
logging
350
amqp
250
license
400
language
350
cronjob
300
mongodb / mongodbrs
4000
mongodb_backup
1000
mongodb_admin/express
250
rabbitmq
2000
hc (linux client)
2500
Last updated