From 3a90b5224df686e6efbf78e6b6cbb5333115ea82 Mon Sep 17 00:00:00 2001
From: Daniel J Walsh <dwalsh@redhat.com>
Date: Tue, 19 Jun 2018 07:50:30 -0400
Subject: Returning joining namespace error should not be fatal

I got my database state in a bad way by killing a hanging container.

It did not setup the network namespace correctly

listing/remove bad containers becomes impossible.

podman run alpine/nginx
^c
got me in this state.

I got into a state in the database where
podman ps -a
was returning errors and I could not get out of it,  Makeing joining the network
namespace a non fatal error fixes the issue.

Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>

Closes: #918
Approved by: mheon
---
 libpod/boltdb_state.go          |  8 +++++---
 libpod/boltdb_state_internal.go | 11 +++++++----
 libpod/runtime_ctr.go           |  8 +++++---
 3 files changed, 17 insertions(+), 10 deletions(-)

diff --git a/libpod/boltdb_state.go b/libpod/boltdb_state.go
index 3ee24122a..d7c950fa0 100644
--- a/libpod/boltdb_state.go
+++ b/libpod/boltdb_state.go
@@ -455,10 +455,12 @@ func (s *BoltState) UpdateContainer(ctr *Container) error {
 
 			// Open the new network namespace
 			ns, err := joinNetNS(netNSPath)
-			if err != nil {
-				return errors.Wrapf(err, "error joining network namespace for container %s", ctr.ID())
+			if err == nil {
+				newState.NetNS = ns
+			} else {
+				logrus.Errorf("error joining network namespace for container %s", ctr.ID())
+				ctr.valid = false
 			}
-			newState.NetNS = ns
 		}
 	} else {
 		// The container no longer has a network namespace
diff --git a/libpod/boltdb_state_internal.go b/libpod/boltdb_state_internal.go
index 407d2758a..44f764be1 100644
--- a/libpod/boltdb_state_internal.go
+++ b/libpod/boltdb_state_internal.go
@@ -200,6 +200,7 @@ func getRuntimeConfigBucket(tx *bolt.Tx) (*bolt.Bucket, error) {
 }
 
 func (s *BoltState) getContainerFromDB(id []byte, ctr *Container, ctrsBkt *bolt.Bucket) error {
+	valid := true
 	ctrBkt := ctrsBkt.Bucket(id)
 	if ctrBkt == nil {
 		return errors.Wrapf(ErrNoSuchCtr, "container %s not found in DB", string(id))
@@ -230,10 +231,12 @@ func (s *BoltState) getContainerFromDB(id []byte, ctr *Container, ctrsBkt *bolt.
 	if netNSBytes != nil {
 		nsPath := string(netNSBytes)
 		netNS, err := joinNetNS(nsPath)
-		if err != nil {
-			return errors.Wrapf(err, "error joining network namespace for container %s", string(id))
+		if err == nil {
+			ctr.state.NetNS = netNS
+		} else {
+			logrus.Errorf("error joining network namespace for container %s", ctr.ID())
+			valid = false
 		}
-		ctr.state.NetNS = netNS
 	}
 
 	// Get the lock
@@ -245,7 +248,7 @@ func (s *BoltState) getContainerFromDB(id []byte, ctr *Container, ctrsBkt *bolt.
 	ctr.lock = lock
 
 	ctr.runtime = s.runtime
-	ctr.valid = true
+	ctr.valid = valid
 
 	return nil
 }
diff --git a/libpod/runtime_ctr.go b/libpod/runtime_ctr.go
index 932bebdef..82a2fed19 100644
--- a/libpod/runtime_ctr.go
+++ b/libpod/runtime_ctr.go
@@ -165,9 +165,11 @@ func (r *Runtime) RemoveContainer(ctx context.Context, c *Container, force bool)
 // Locks the container, but does not lock the runtime
 func (r *Runtime) removeContainer(ctx context.Context, c *Container, force bool) error {
 	if !c.valid {
-		// Container probably already removed
-		// Or was never in the runtime to begin with
-		return nil
+		if ok, _ := r.HasContainer(c.ID()); !ok {
+			// Container probably already removed
+			// Or was never in the runtime to begin with
+			return nil
+		}
 	}
 
 	// We need to lock the pod before we lock the container
-- 
cgit v1.2.3-54-g00ecf