From 5949405d0031a4aba91c81cb28c24821ad2d439a Mon Sep 17 00:00:00 2001
From: Reid Wahl <nwahl@redhat.com>
Date: Thu, 3 Jan 2019 15:05:20 -0800
Subject: [PATCH] docker: Fix issues with stop operation

The docker RA's stop operation doesn't behave properly in some cases.
  1. It returns a false success code in case of an error response from
     the daemon.
  2. It fails at `remove_container()` if the container does not exist
     but another docker object of the same name does exist.

In case #1, the `container_exists()` function returns the same exit code
(1) if the container is not found (an expected error) or if there is an
error response from the docker daemon (an unexpected error). These types
of errors should be handled differently.

In case #2, the `docker inspect` calls do not limit their search to
containers. So if a non-container object is found with a matching name,
the RA attempts to remove a container by that name. Such a container may
not exist.

This patch fixes these issues as follows:
  1. Match an error response in `container_exists()` against the string
     "No such container".
  2. Add `--type=container` to the `docker inspect` calls to restrict
     the match.
---
 heartbeat/docker | 26 ++++++++++++++++++++++----
 1 file changed, 22 insertions(+), 4 deletions(-)

diff --git a/heartbeat/docker b/heartbeat/docker
index f5ba83ff2..c206344ad 100755
--- a/heartbeat/docker
+++ b/heartbeat/docker
@@ -215,7 +215,7 @@ monitor_cmd_exec()
 		out=$(docker exec ${CONTAINER} $OCF_RESKEY_monitor_cmd 2>&1)
 		rc=$?
 	else
-		out=$(echo "$OCF_RESKEY_monitor_cmd" | nsenter --target $(docker inspect --format {{.State.Pid}} ${CONTAINER}) --mount --uts --ipc --net --pid 2>&1)
+		out=$(echo "$OCF_RESKEY_monitor_cmd" | nsenter --target $(docker inspect --type=container --format {{.State.Pid}} ${CONTAINER}) --mount --uts --ipc --net --pid 2>&1)
 		rc=$?
 	fi
 
@@ -236,7 +236,25 @@ monitor_cmd_exec()
 
 container_exists()
 {
-	docker inspect --format {{.State.Running}} $CONTAINER | egrep '(true|false)' >/dev/null 2>&1
+	local err
+
+	err=$(docker inspect --type=container $CONTAINER 2>&1 >/dev/null)
+
+	if [ $? -ne $OCF_SUCCESS ]; then
+		case $err in
+			*"No such container"*)
+				# Return failure instead of exiting if container does not exist
+				return 1
+				;;
+			*)
+				# Exit if error running command
+				ocf_exit_reason "$err"
+				exit $OCF_ERR_GENERIC
+				;;
+		esac
+	fi
+
+	return $OCF_SUCCESS
 }
 
 remove_container()
@@ -265,7 +283,7 @@ docker_simple_status()
 	fi
 
 	# retrieve the 'Running' attribute for the container
-	val=$(docker inspect --format {{.State.Running}} $CONTAINER 2>/dev/null)
+	val=$(docker inspect --type=container --format {{.State.Running}} $CONTAINER 2>/dev/null)
 	if [ $? -ne 0 ]; then
 		#not running as a result of container not being found
 		return $OCF_NOT_RUNNING
@@ -295,7 +313,7 @@ docker_health_status()
                 # if starting takes longer than monitor timeout then upstream will make this fail.
                 while
 
-                        val=$(docker inspect --format {{.State.Health.Status}} $CONTAINER 2>/dev/null)
+                        val=$(docker inspect --type=container --format {{.State.Health.Status}} $CONTAINER 2>/dev/null)
                         if [ $? -ne 0 ]; then
                                 #not healthy as a result of container not being found
                                 return $OCF_NOT_RUNNING