From 11c11b06dc686bb0e8f0cd7de9aa53dbcc6498b0 Mon Sep 17 00:00:00 2001
From: "A.Unger" <zyxancf@gmail.com>
Date: Fri, 23 Apr 2021 14:13:58 +0200
Subject: [PATCH 1/3] stop supervisor if a service fails to start

---
 ocis/pkg/runtime/service/service.go | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/ocis/pkg/runtime/service/service.go b/ocis/pkg/runtime/service/service.go
index cfab669058..464df996cf 100644
--- a/ocis/pkg/runtime/service/service.go
+++ b/ocis/pkg/runtime/service/service.go
@@ -123,6 +123,10 @@ func Start(o ...Option) error {
 		return err
 	}
 
+	// halt listens for interrupt signals and blocks.
+	halt := make(chan os.Signal, 1)
+	signal.Notify(halt, syscall.SIGINT, syscall.SIGTERM, syscall.SIGQUIT, syscall.SIGHUP)
+
 	// notify goroutines that they are running on supervised mode
 	s.cfg.Mode = ociscfg.SUPERVISED
 
@@ -131,6 +135,9 @@ func Start(o ...Option) error {
 	// Start creates its own supervisor. Running services under `ocis server` will create its own supervision tree.
 	s.Supervisor = suture.New("ocis", suture.Spec{
 		EventHook: func(e suture.Event) {
+			if e.Type() == suture.EventTypeServiceTerminate {
+				halt <- os.Interrupt
+			}
 			s.Log.Info().Str("event", e.String()).Msg(fmt.Sprintf("supervisor: %v", e.Map()["supervisor_name"]))
 		},
 	})
@@ -148,10 +155,6 @@ func Start(o ...Option) error {
 	}
 	rpc.HandleHTTP()
 
-	// halt listens for interrupt signals and blocks.
-	halt := make(chan os.Signal, 1)
-	signal.Notify(halt, syscall.SIGINT, syscall.SIGTERM, syscall.SIGQUIT, syscall.SIGHUP)
-
 	l, err := net.Listen("tcp", net.JoinHostPort(s.cfg.Runtime.Host, s.cfg.Runtime.Port))
 	if err != nil {
 		s.Log.Fatal().Err(err)

From 8da06d32b95a717a34cf5fddbbec25cde49ab40d Mon Sep 17 00:00:00 2001
From: "A.Unger" <zyxancf@gmail.com>
Date: Fri, 23 Apr 2021 14:15:30 +0200
Subject: [PATCH 2/3] add changelog

---
 changelog/unreleased/stop-supervisor-if-failure.md | 7 +++++++
 1 file changed, 7 insertions(+)
 create mode 100644 changelog/unreleased/stop-supervisor-if-failure.md

diff --git a/changelog/unreleased/stop-supervisor-if-failure.md b/changelog/unreleased/stop-supervisor-if-failure.md
new file mode 100644
index 0000000000..8bccf7a53d
--- /dev/null
+++ b/changelog/unreleased/stop-supervisor-if-failure.md
@@ -0,0 +1,7 @@
+Bugfix: Stop the supervisor if a service fails to start
+
+Steps to make the supervisor fail:
+
+`PROXY_HTTP_ADDR=0.0.0.0:9144 bin/ocis server`
+
+https://github.com/owncloud/ocis/pull/1963

From 420447009bac7e6e491d6f5d40c2deb3550d11fa Mon Sep 17 00:00:00 2001
From: "A.Unger" <zyxancf@gmail.com>
Date: Fri, 23 Apr 2021 15:19:03 +0200
Subject: [PATCH 3/3] tweak backoff tolerance

---
 ocis/pkg/runtime/service/service.go | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/ocis/pkg/runtime/service/service.go b/ocis/pkg/runtime/service/service.go
index 464df996cf..268d7473fd 100644
--- a/ocis/pkg/runtime/service/service.go
+++ b/ocis/pkg/runtime/service/service.go
@@ -132,14 +132,23 @@ func Start(o ...Option) error {
 
 	setMicroLogger()
 
+	// tolerance controls backoff cycles from the supervisor.
+	tolerance := 5
+	totalBackoff := 0
+
 	// Start creates its own supervisor. Running services under `ocis server` will create its own supervision tree.
 	s.Supervisor = suture.New("ocis", suture.Spec{
 		EventHook: func(e suture.Event) {
-			if e.Type() == suture.EventTypeServiceTerminate {
-				halt <- os.Interrupt
+			if e.Type() == suture.EventTypeBackoff {
+				totalBackoff++
+				if totalBackoff == tolerance {
+					halt <- os.Interrupt
+				}
 			}
 			s.Log.Info().Str("event", e.String()).Msg(fmt.Sprintf("supervisor: %v", e.Map()["supervisor_name"]))
 		},
+		FailureThreshold: 5,
+		FailureBackoff:   3 * time.Second,
 	})
 
 	// reva storages have their own logging. For consistency sake the top level logging will cascade to reva.