Is your feature request related to a problem? Please describe.
The in_tcp, in_udp, and in_forward plugins don't support the workers configuration that HTTP-based plugins have. This means these plugins can only use one thread to handle incoming traffic, which can become a bottleneck when processing high volumes of data.
HTTP plugins like in_http, in_splunk, and in_opentelemetry already support workers (recently added). This allows them to use multiple threads and better utilize CPU cores.
Describe the solution you'd like
Add a workers option to in_tcp, in_udp, and in_forward plugins, similar to what exists for HTTP plugins.
This would:
- Allow users configure multiple worker threads to handle connections.
- Better utilize multi-core CPUs when handling high traffic.
- Keep the current behavior by default (workers=1).
Describe alternatives you've considered
- Running multiple Fluent Bit instances with different ports (or scale in k8s deployments) - adds complexity and needs external load balancing.
Is your feature request related to a problem? Please describe.
The
in_tcp,in_udp, andin_forwardplugins don't support theworkersconfiguration that HTTP-based plugins have. This means these plugins can only use one thread to handle incoming traffic, which can become a bottleneck when processing high volumes of data.HTTP plugins like
in_http,in_splunk, andin_opentelemetryalready support workers (recently added). This allows them to use multiple threads and better utilize CPU cores.Describe the solution you'd like
Add a
workersoption toin_tcp,in_udp, andin_forwardplugins, similar to what exists for HTTP plugins.This would:
Describe alternatives you've considered