Skip to content

docker-backend-overview.md

Latest commit

 

History

History
34 lines (26 loc) · 2.52 KB

docker-backend-overview.md

File metadata and controls

34 lines (26 loc) · 2.52 KB

Docker backend specifics

Overview

SCT supports multiple backends for deploying and testing Scylla clusters, with the Docker backend being a practical choice for local development and testing.
That said, there are a few differences and specifics to be aware of when using the Docker backend, compared to other backends like AWS, GCE, or Azure.

Running Nemeses

Not all nemeses are supported or run successfully on the Docker backend.
The full list of supported, unsupported or failing nemeses (due to know issues) can be found on the Individual Nemesis status on Docker backend page.

Monitoring stack is on the Docker host machine

SCT does not support creating a dedicated monitoring node when using the Docker backend. As a result, the monitoring stack is installed directly on the host machine, not on a dedicated Docker instance.

Scylla-manager is not installed

SCT installs Scylla-manager on a monitoring node, which is not supported in the Docker backend. Therefore, Scylla-manager is not installed when using this backend.

Starting DB node instances with specific resources footprint

By default, Scylla on containerized DB instances is started with the following CPU and RAM configuration:

  • smp parameter is set to 1
  • memory parameter is not set

The default values are sufficient for simple test configurations. But for more complex scenarios the following issues may arise:

  • if the memory parameter is not set or does not limit the memory usage per instance, RAM of the local machine can be exhausted (especially with multiple DB instances in the test configuration), causing the test to fail
  • if the default smp parameter value is used, DB instance containers may become overloaded during stress commands execution, leading to test failures

To prevent these issues, the smp and memory parameters should be set according to the configuration and/or load profile of the test scenario.
This can be done by setting the appropriate values through the append_scylla_args SCT config parameter in the test configuration file. For example:

append_scylla_args: '--smp 2 --memory 2G'

Executing a longevity test on the Docker backend in Jenkins

SCT longevity tests can be executed on the Docker backend in Jenkins, with AWS serving as the cloud provider for SCT runner instance.
This setup allows simulating test execution as if it was running on a local machine. The SCT runner instance in AWS operates as an all-in-one setup, hosting the loaders, DB nodes and monitoring stack.