Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add pprof endpoints if enabled in configuration #1017

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add pprof endpoints if enabled in configuration #1017

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions cmd/kured/main.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,7 @@ var (
period time.Duration
metricsHost string
metricsPort int
pprofEnabled bool
drainGracePeriod int
drainPodSelector string
skipWaitForDeleteTimeoutSeconds int
Expand Down Expand Up @@ -120,6 +121,8 @@ func main() {
"host where metrics will listen")
flag.IntVar(&metricsPort, "metrics-port", 8080,
"port number where metrics will listen")
flag.BoolVar(&pprofEnabled, "pprof-enabled", false,
"enable /debug/pprof on metrics port")
flag.IntVar(&drainGracePeriod, "drain-grace-period", -1,
"time in seconds given to each pod to terminate gracefully, if negative, the default value specified in the pod will be used")
flag.StringVar(&drainPodSelector, "drain-pod-selector", "",
Expand Down Expand Up @@ -278,6 +281,9 @@ func main() {
go rebootAsRequired(nodeID, rebooter, rebootChecker, window, lock, client)
go maintainRebootRequiredMetric(nodeID, rebootChecker)

if pprofEnabled {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this works: while pprofEnabled is by defaut to false, the net/http/pprof is still loaded, so it's still injecting itself?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh yeah.. 🤦

i guess we could just import _ "net/http/pprof" and leave it enabled.
Or use a custom mux and not the DefaultServeMux where we can conditionally enable it.

Which do you like?

Copy link
Collaborator

@evrardjp evrardjp Nov 7, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't know yet... Do we really need to have this enabled at all times? For me we do not... And we do not have to flip the switch live on a build. I don't mind continuing to add import _ "net/http/pprof" line when necessary.

For me the valid question is not about whether we do have profiling, but rather "who is checking the profiling before release". yet I wouldn't mind to make the life easier of the person who will need it. Just not sure how yet.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pprof is normally enabled on all services and its because all kind of leaks can be hard to reproduce and then we can just check the pprof endpoint in production when needed.

I'll update the PR to default to off with another mux and then we can in our deployment always enable it like almost all our other pods.

Copy link
Collaborator

@evrardjp evrardjp Nov 7, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pprof is normally enabled on all services and its because all kind of leaks can be hard to reproduce and then we can just check the pprof endpoint in production when needed.

If I follow your point, we would just add import _ "net/http/pprof" then. However I don't think it's a good idea, it's been the cause of cves in the past. GHSA-9frv-h2cf-52wh

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah that's why I would like it default off but on for all our hundreds of instances.

enablePPROF()
}
http.Handle("/metrics", promhttp.Handler())
log.Fatal(http.ListenAndServe(fmt.Sprintf("%s:%d", metricsHost, metricsPort), nil))
}
Expand Down
15 changes: 15 additions & 0 deletions cmd/kured/pprof.go
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,15 @@
package main

import (
"net/http"
"net/http/pprof"
)

func enablePPROF() {
prefix := "GET "
http.HandleFunc(prefix+"/debug/pprof/", pprof.Index)
http.HandleFunc(prefix+"/debug/pprof/cmdline", pprof.Cmdline)
http.HandleFunc(prefix+"/debug/pprof/profile", pprof.Profile)
http.HandleFunc(prefix+"/debug/pprof/symbol", pprof.Symbol)
http.HandleFunc(prefix+"/debug/pprof/trace", pprof.Trace)
}