Skip to content

Commit

Permalink
Merge pull request #13 from FortnoxAB/reapzombies
Browse files Browse the repository at this point in the history
Only reap zombies after each exec.
  • Loading branch information
jonaz authored Nov 29, 2024
2 parents 8b73301 + 92b52b6 commit a433fd2
Show file tree
Hide file tree
Showing 4 changed files with 37 additions and 33 deletions.
3 changes: 2 additions & 1 deletion docker-compose.yml
Original file line number Diff line number Diff line change
Expand Up @@ -59,11 +59,12 @@ services:

agent-1:
image: renovate/renovate:39.21.1-full
command:
entrypoint:
- /renovator/renovator
- --loglevel=debug
- agent
- --redis-url=redis://redis:6379/0
# - --max-process-count=5
depends_on:
init-agent-1:
condition: service_completed_successfully
Expand Down
28 changes: 0 additions & 28 deletions pkg/agent/agent.go
Original file line number Diff line number Diff line change
Expand Up @@ -2,13 +2,9 @@ package agent

import (
"context"
"errors"
"fmt"
"os"
"os/signal"
"strings"
"sync"
"syscall"
"time"

"github.com/fortnoxab/renovator/pkg/command"
Expand Down Expand Up @@ -52,7 +48,6 @@ func NewAgentFromContext(cCtx *cli.Context) (*Agent, error) {
}

func (a *Agent) Run(ctx context.Context) {
ZombieReaper()
reposToProcess := make(chan string)

wg := &sync.WaitGroup{}
Expand Down Expand Up @@ -108,26 +103,3 @@ func (a *Agent) Run(ctx context.Context) {
}
wg.Wait()
}

func ZombieReaper() {
signals := make(chan os.Signal, 1)
signal.Notify(signals, syscall.SIGCHLD)

go func() {
for range signals {
for {
var wstatus syscall.WaitStatus
time.Sleep(1 * time.Second)
pid, err := syscall.Wait4(-1, &wstatus, 0, nil)
if errors.Is(err, syscall.ECHILD) {
break
}
if err != nil {
logrus.Errorf("error waiting for child %d: %s", pid, err)
continue
}
logrus.Debugf("reaped zombie %d %d", pid, wstatus)
}
}
}()
}
37 changes: 35 additions & 2 deletions pkg/command/run.go
Original file line number Diff line number Diff line change
@@ -1,8 +1,13 @@
package command

import (
"errors"
"os"
"os/exec"
"strings"
"syscall"

"github.com/sirupsen/logrus"
)

type Commander interface {
Expand All @@ -24,6 +29,34 @@ func (e *Exec) RunWithEnv(env []string, head string, parts ...string) (err error
cmd.Env = os.Environ()
cmd.Env = append(cmd.Env, env...)

err = cmd.Run()
return
err = cmd.Start()
if err != nil {
return err
}

logrus.Debugf("started pid %d, %s", cmd.Process.Pid, head+" "+strings.Join(parts, " "))

err = cmd.Wait()

e.reapChildren()
return err
}

func (e *Exec) reapChildren() {
for {
var wstatus syscall.WaitStatus
pid, err := syscall.Wait4(-1, &wstatus, syscall.WNOHANG, nil)
if errors.Is(err, syscall.ECHILD) {
return
}
if err != nil {
logrus.Errorf("error waiting for child %d: %s", pid, err)
continue
}
if pid == 0 && wstatus == 0 {
logrus.Debug("found no pid")
return
}
logrus.Debugf("reaped zombie %d %d", pid, wstatus)
}
}
2 changes: 0 additions & 2 deletions pkg/master/master.go
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,6 @@ import (
"fmt"
"sync"

"github.com/fortnoxab/renovator/pkg/agent"
"github.com/fortnoxab/renovator/pkg/command"
"github.com/fortnoxab/renovator/pkg/kafka"
"github.com/fortnoxab/renovator/pkg/leaderelect"
Expand Down Expand Up @@ -67,7 +66,6 @@ func (m *Master) Run(ctx context.Context) error {
if m.CronSchedule == nil {
return doRun(ctx, m.Candidate, m.RedisClient, m.Renovator, m.LeaderElect)
}
agent.ZombieReaper()

if m.RunFirstTime {
logrus.Info("running due to --run-first-time")
Expand Down

0 comments on commit a433fd2

Please sign in to comment.