feat(cmd/debuginfo) add new metrics to be collected

[OmarAyo]

• This PR adds 5 new metrics (/jemalloc, /state, /health, /debug/vars, /metrics) to be collected when running the debuginfo
• the flag -p has been changed to -m to reflect the change. This new flag will let you pick up one or multiple metrics/pprof to be collected:

-m, --metrics strings    List of metrics & profile to dump in the report. (default [jemalloc,state,health,vars,metrics,heap,cpu_profile,trace,goroutine,threadcreate,block,mutex])

• default value of -s flag is now 30sec since it requires 30s to collect a cpu profile
• when saving the metric/profile - it will log the metric/profile name and the path where it's saving to.
• added new flag for time based profiles that are cpu and trace profiles:

-c, --cron_pprof strings   time-based pprof (default [cpu_profile,trace])

output of running the debuginfo command:

 dgraph debuginfo -s 30
I0305 21:01:44.730085   10587 run.go:126] using directory /tmp/dgraph-debuginfo041885197 for debug info dump.
I0305 21:01:44.730251   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/jemalloc
I0305 21:01:44.730261   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.731077   10587 debugging.go:51] saving jemalloc metric in /tmp/dgraph-debuginfo041885197/alpha_jemalloc.gz
I0305 21:01:44.731086   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/state
I0305 21:01:44.731092   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.731720   10587 debugging.go:51] saving state metric in /tmp/dgraph-debuginfo041885197/alpha_state.gz
I0305 21:01:44.731731   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/health
I0305 21:01:44.731736   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.732048   10587 debugging.go:51] saving health metric in /tmp/dgraph-debuginfo041885197/alpha_health.gz
I0305 21:01:44.732058   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/vars
I0305 21:01:44.732065   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.732557   10587 debugging.go:51] saving vars metric in /tmp/dgraph-debuginfo041885197/alpha_vars.gz
I0305 21:01:44.732568   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/metrics
I0305 21:01:44.732573   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.734904   10587 debugging.go:51] saving metrics metric in /tmp/dgraph-debuginfo041885197/alpha_metrics.gz
I0305 21:01:44.734912   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/heap
I0305 21:01:44.734917   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.738036   10587 debugging.go:51] saving heap metric in /tmp/dgraph-debuginfo041885197/alpha_heap.gz
I0305 21:01:44.738048   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/goroutine?debug=2
I0305 21:01:44.738057   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.739136   10587 debugging.go:51] saving goroutine metric in /tmp/dgraph-debuginfo041885197/alpha_goroutine.gz
I0305 21:01:44.739145   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/threadcreate
I0305 21:01:44.739151   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.740182   10587 debugging.go:51] saving threadcreate metric in /tmp/dgraph-debuginfo041885197/alpha_threadcreate.gz
I0305 21:01:44.740192   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/block
I0305 21:01:44.740198   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.741154   10587 debugging.go:51] saving block metric in /tmp/dgraph-debuginfo041885197/alpha_block.gz
I0305 21:01:44.741163   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/mutex
I0305 21:01:44.741169   10587 debugging.go:76] please wait... (30s)
I0305 21:01:44.742330   10587 debugging.go:51] saving mutex metric in /tmp/dgraph-debuginfo041885197/alpha_mutex.gz
I0305 21:01:44.742341   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/profile?seconds=30s
I0305 21:01:44.742348   10587 debugging.go:76] please wait... (30s)
I0305 21:02:14.812492   10587 debugging.go:63] saving cpu_profile metric in /tmp/dgraph-debuginfo041885197/alpha_cpu_profile.gz
I0305 21:02:14.812570   10587 debugging.go:74] fetching information over HTTP from http://localhost:8080/debug/pprof/trace?seconds=30s
I0305 21:02:14.812596   10587 debugging.go:76] please wait... (30s)
I0305 21:02:15.816449   10587 debugging.go:63] saving trace metric in /tmp/dgraph-debuginfo041885197/alpha_trace.gz
I0305 21:02:15.842375   10587 run.go:159] Debuginfo archive successful: dgraph-debuginfo041885197.tar.gz

---

This change is 

[danielmai]

Reviewable status: 0 of 2 files reviewed, 1 unresolved discussion (waiting on @manishrjain, @OmarAyo, and @vvbalaji-dgraph)

---

dgraph/cmd/debuginfo/debugging.go, line 42 at r1 (raw file):

}

var metricTypes = []string{

This can be a struct for the metric types. It can have a name and its endpoint. Something like

type debugMetric struct {
	name     string
	endpoint string
}

var metricTypes = []debugMetric{
	{
		name:     "jemalloc",
		endpoint: "/jemalloc",
	},
	{
		name:     "state",
		endpoint: "/state",
	},
	{
		name:     "health",
		endpoint: "/health",
	},
	{
		name:     "vars",
		endpoint: "/debug/vars",
	},
	{
		name:     "metrics",
		endpoint: "/metrics",
	},
}

Then the name and endpoint can be used separately.

[OmarAyo]

---

dgraph/cmd/debuginfo/debugging.go, line 42 at r1 (raw file):

Previously, danielmai (Daniel Mai) wrote…

This can be a struct for the metric types. It can have a name and its endpoint. Something like

type debugMetric struct {
	name     string
	endpoint string
}

var metricTypes = []debugMetric{
	{
		name:     "jemalloc",
		endpoint: "/jemalloc",
	},
	{
		name:     "state",
		endpoint: "/state",
	},
	{
		name:     "health",
		endpoint: "/health",
	},
	{
		name:     "vars",
		endpoint: "/debug/vars",
	},
	{
		name:     "metrics",
		endpoint: "/metrics",
	},
}

Then the name and endpoint can be used separately.

Addressed this by using a map instead

[danielmai]

Reviewed 1 of 2 files at r2.
Reviewable status: 1 of 2 files reviewed, 5 unresolved discussions (waiting on @danielmai, @manishrjain, @OmarAyo, and @vvbalaji-dgraph)

---

dgraph/cmd/debuginfo/run.go, line 55 at r2 (raw file):

	"cpu_profile":  "/debug/pprof/profile",
	"trace":        "/debug/pprof/trace",
	"goroutine":    "/debug/pprof/goroutine",

This should be /debug/pprof/goroutine?debug=2 for the right goroutine stack format: https://dgraph.io/docs/howto/retrieving-debug-information/#goroutine-stack

---

dgraph/cmd/debuginfo/debugging.go, line 55 at r2 (raw file):

the debug

the debug info

---

dgraph/cmd/debuginfo/debugging.go, line 75 at r2 (raw file):

dump file

debug file

---

dgraph/cmd/debuginfo/pprof.go, line 53 at r2 (raw file):

	for _, profileType := range profiles {
		source := fmt.Sprintf("%s/debug/pprof/%s?duration=%d", u.String(),

Are we still setting ?duration in the new one?

[OmarAyo]

---

dgraph/cmd/debuginfo/debugging.go, line 42 at r1 (raw file):

Previously, OmarAyo (OmarAyo) wrote…

Addressed this by using a map instead

Done.

[OmarAyo]

---

dgraph/cmd/debuginfo/pprof.go, line 53 at r2 (raw file):

Previously, danielmai (Daniel Mai) wrote…

Are we still setting ?duration in the new one?

Done.

[OmarAyo]

---

dgraph/cmd/debuginfo/debugging.go, line 55 at r2 (raw file):

Previously, danielmai (Daniel Mai) wrote…

the debug

the debug info

Done.

[OmarAyo]

---

dgraph/cmd/debuginfo/debugging.go, line 75 at r2 (raw file):

Previously, danielmai (Daniel Mai) wrote…

dump file

debug file

Done.

[OmarAyo]

---

dgraph/cmd/debuginfo/run.go, line 55 at r2 (raw file):

Previously, danielmai (Daniel Mai) wrote…

This should be /debug/pprof/goroutine?debug=2 for the right goroutine stack format: https://dgraph.io/docs/howto/retrieving-debug-information/#goroutine-stack

Done.

[OmarAyo]

Reviewed 1 of 2 files at r4, 1 of 1 files at r5.
Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @danielmai, @manishrjain, and @vvbalaji-dgraph)