Improve server performance through Synapse workers

[psrpinto]

Currently the Matrix server (homeserver) of community.wordpress.org is running as a single process. However, it can be run as separate processes, which should improve performance. From Synapse's documentation on workers:

For larger instances where performance is a concern it can be helpful to split out functionality into multiple separate python processes. These processes are called 'workers', and are (eventually) intended to scale horizontally independently.

The architecture looks like the following (from How we fixed Synapse's scalability):

References

• https://github.com/matrix-org/synapse/tree/develop/contrib/docker_compose_workers
• https://github.com/matrix-org/synapse/tree/develop/contrib/workers-bash-scripts

Metrics

"Event send time" should be a good indicator of overall performance.

Before

After

TODO