Installation requirements
Operating system
Installation of the on-premise version is possible on GNU/Linux distributions.
Server resources
Baseline requirements for production installation
- Minimum supported resources for application servers: 32Gb RAM, 16 CPU cores.
For production installations, it is not recommended to go below these values, as this may lead to performance degradation and unstable platform operation.
Use a CPU compatible with the Intel Scalable Gen 2 family. CodeScoring requires CPU instructions that may not be present in older CPUs.
Data services requirements
- PostgreSQL: from 32Gb RAM;
- Redis: from 2Gb RAM;
shmsize for PostgreSQL: at least 4Gb;- If an external database is used, PostgreSQL is recommended to have at least 64Gb RAM.
Storage requirements
- For CodeScoring.SCA with VCS projects, calculate
analysis-rootvolume based on the size of analyzed repositories multiplied by three; - For CodeScoring.SCA with CLI projects and for CodeScoring.OSA, there is no mandatory
×3storage formula; - For the offline installation allocate at least 300Gb for the CodeScoring Index database, with additional capacity reserved for further updates.
Example of a high-load installation
- Database server: 96Gb RAM, 24 CPU cores;
- Application server: 192Gb RAM, 48 CPU cores.
Supported versions of external services
When using your own database instances, make sure that their versions meet the requirements below:
Redis
- Minimum version: 7.0.0;
- Tested version: 7.4.9.
PostgreSQL
- Minimum version: 15.x (any minor);
- Tested version: 15.15.
Using other major branches does not guarantee correct results and may lead to errors or performance degradation.
External requests
To install the system, the Docker Registry with CodeScoring container images must be available, the address of which is provided along with the activation key.
For correct operation, the address index.codescoring.ru with a constantly updated database of known packages must also be available.
From the Index API the platform receives additional information on the dependencies found, their licenses and vulnerabilities.
The general architecture of the work is shown in the image below.

The source code is not transferred from the platform to the CodeScoring cloud. However, to obtain information on dependencies and control licensing terms, the following is sent:
- anonymized information on found package manager manifests and their contents;
- hashes of source code files to search for direct inclusions of Open Source libraries in project code;
- number of active authors over the past year;
- number of projects in the system;
- platform version.
The paths of the manifests and the names of the hashed files are specifically anonymized. The hashes of files whose size does not exceed 512 bytes are not being sent from the platform to the cloud.
An example of a request from an platform to the Index API with data from package manager manifests:
An example of a request from the platform to the Index API with data on hashes of source code files:
