Skip to content

Latest commit

 

History

History
105 lines (85 loc) · 3.56 KB

README.md

File metadata and controls

105 lines (85 loc) · 3.56 KB

Continuous Profiling for Caddy

The package contains 3 Caddy modules for push-mode continuous profiling support in Caddy.

Go Reference

Modules

profiling

The module can only serve as an app that is the host for multiple other profilers to configure and propagate common profiling parameters, i.e. CPU profile rate, block profile rate, mutex profile fraction, and the enabled profile types.

Documentation: https://caddyserver.com/docs/json/apps/profiling/

Sample config

{
	"apps": {
		"profiling": {
			"cpu_profile_rate": 2,
			"block_profile_rate": 2,
			"mutex_profile_fraction": 2,
			"profile_types": ["cpu","heap","allocs","goroutine"],
			"profilers": [
				{
					"profiler": "pyroscope",
					"server_address": "http://localhost:4040"
				},
				{
					"profiler": "pyroscope",
					"server_address": "http://another-instance.local:4040",
					"profile_types": ["heap"]
				},
				{
					"profiler": "profefe",
					// rest of config
				}
			]
		}
	}
}

pyroscope

Configures and enables the push-mode Pyroscope agent. May serve as an app or a child module of the profiling app. It may be configured as a child profiler of the profilig app or as first-level app within Caddy. Configuring the pyroscope app may look like this:

{
	"apps": {
		"pyroscope": {
			"server_address": "http://localhost:4040",
			"application_name": "my_cool_app",
			"auth_token": "{env.PYROSCOPE_AUTH_TOKEN}",
			"parameters": {
				"cpu_profile_rate": 2,
				"block_profile_rate": 2,
				"mutex_profile_fraction": 2,
				"profile_types": ["cpu","heap","allocs","goroutine"]
			}
		}
	}
}

Documentation: https://caddyserver.com/docs/json/apps/pyroscope/

profefe

Similar to the pyroscope module, the profefe module configures and pushes data to Profefe server. May serve as an app or a child module of the profiling app. It may be configured as a child profiler of the profilig app or as first-level app within Caddy. Configuring the profefe app may look like this:

{
	"apps": {
		"profefe": {
			"address": "http://localhost:4040",
			"service": "my_cool_app",
			"timeout": "10m",
			"parameters": {
				"cpu_profile_rate": 2,
				"block_profile_rate": 2,
				"mutex_profile_fraction": 2,
				"profile_types": ["cpu","heap","allocs","goroutine"]
			}
		}
	}
}

Documentation: https://caddyserver.com/docs/json/apps/profefe/

Available Profile Types

The ProfileTypes field in Parameters accepts the following values:

  • goroutine: stack traces of all current goroutines
  • heap: a sampling of memory allocations of live objects
  • allocs: a sampling of all past memory allocations
  • threadcreate: stack traces that led to the creation of new OS threads
  • block: stack traces that led to blocking on synchronization primitives
  • mutex: stack traces of holders of contended mutexes

They may be overridden by the individual guest profilers, if the guest module chooses to do so. The guest modules may not support the same types and they may have different names for the same types. The guest profilers own the translation logic.