diff options
author | Fabricio Voznika <fvoznika@google.com> | 2020-11-05 15:34:24 -0800 |
---|---|---|
committer | gVisor bot <gvisor-bot@google.com> | 2020-11-05 15:36:20 -0800 |
commit | c47f8afe2334c1395a2acd9ebf5191aa4554907c (patch) | |
tree | 10eccac0af9efe876167cf070396d0c6068c44ec /runsc/boot | |
parent | caf1e52e18222d9eace348fa53e6857c78e15479 (diff) |
Fix failure setting OOM score adjustment
When OOM score adjustment needs to be set, all the containers need to be
loaded to find all containers that belong to the sandbox. However, each
load signals the container to ensure it is still alive. OOM score
adjustment is set during creation and deletion of every container, generating
a flood of signals to all containers. The fix removes the signal check
when it's not needed.
There is also a race fetching OOM score adjustment value from the parent when
the sandbox exits at the same time (the time it took to signal containers above
made this window quite large). The fix is to store the original value
in the sandbox state file and use it when the value needs to be restored.
Also add more logging and made the existing ones more consistent to help with
debugging.
PiperOrigin-RevId: 340940799
Diffstat (limited to 'runsc/boot')
-rw-r--r-- | runsc/boot/controller.go | 31 | ||||
-rw-r--r-- | runsc/boot/loader.go | 2 |
2 files changed, 17 insertions, 16 deletions
diff --git a/runsc/boot/controller.go b/runsc/boot/controller.go index 4e0f0d57a..fdf13c8e1 100644 --- a/runsc/boot/controller.go +++ b/runsc/boot/controller.go @@ -196,7 +196,7 @@ type containerManager struct { // StartRoot will start the root container process. func (cm *containerManager) StartRoot(cid *string, _ *struct{}) error { - log.Debugf("containerManager.StartRoot %q", *cid) + log.Debugf("containerManager.StartRoot, cid: %s", *cid) // Tell the root container to start and wait for the result. cm.startChan <- struct{}{} if err := <-cm.startResultChan; err != nil { @@ -207,13 +207,13 @@ func (cm *containerManager) StartRoot(cid *string, _ *struct{}) error { // Processes retrieves information about processes running in the sandbox. func (cm *containerManager) Processes(cid *string, out *[]*control.Process) error { - log.Debugf("containerManager.Processes: %q", *cid) + log.Debugf("containerManager.Processes, cid: %s", *cid) return control.Processes(cm.l.k, *cid, out) } // Create creates a container within a sandbox. func (cm *containerManager) Create(cid *string, _ *struct{}) error { - log.Debugf("containerManager.Create: %q", *cid) + log.Debugf("containerManager.Create, cid: %s", *cid) return cm.l.createContainer(*cid) } @@ -237,12 +237,11 @@ type StartArgs struct { // Start runs a created container within a sandbox. func (cm *containerManager) Start(args *StartArgs, _ *struct{}) error { - log.Debugf("containerManager.Start: %+v", args) - // Validate arguments. if args == nil { return errors.New("start missing arguments") } + log.Debugf("containerManager.Start, cid: %s, args: %+v", args.CID, args) if args.Spec == nil { return errors.New("start arguments missing spec") } @@ -269,27 +268,27 @@ func (cm *containerManager) Start(args *StartArgs, _ *struct{}) error { } }() if err := cm.l.startContainer(args.Spec, args.Conf, args.CID, fds); err != nil { - log.Debugf("containerManager.Start failed %q: %+v: %v", args.CID, args, err) + log.Debugf("containerManager.Start failed, cid: %s, args: %+v, err: %v", args.CID, args, err) return err } - log.Debugf("Container %q started", args.CID) + log.Debugf("Container started, cid: %s", args.CID) return nil } // Destroy stops a container if it is still running and cleans up its // filesystem. func (cm *containerManager) Destroy(cid *string, _ *struct{}) error { - log.Debugf("containerManager.destroy %q", *cid) + log.Debugf("containerManager.destroy, cid: %s", *cid) return cm.l.destroyContainer(*cid) } // ExecuteAsync starts running a command on a created or running sandbox. It // returns the PID of the new process. func (cm *containerManager) ExecuteAsync(args *control.ExecArgs, pid *int32) error { - log.Debugf("containerManager.ExecuteAsync: %+v", args) + log.Debugf("containerManager.ExecuteAsync, cid: %s, args: %+v", args.ContainerID, args) tgid, err := cm.l.executeAsync(args) if err != nil { - log.Debugf("containerManager.ExecuteAsync failed: %+v: %v", args, err) + log.Debugf("containerManager.ExecuteAsync failed, cid: %s, args: %+v, err: %v", args.ContainerID, args, err) return err } *pid = int32(tgid) @@ -453,9 +452,9 @@ func (cm *containerManager) Resume(_, _ *struct{}) error { // Wait waits for the init process in the given container. func (cm *containerManager) Wait(cid *string, waitStatus *uint32) error { - log.Debugf("containerManager.Wait") + log.Debugf("containerManager.Wait, cid: %s", *cid) err := cm.l.waitContainer(*cid, waitStatus) - log.Debugf("containerManager.Wait returned, waitStatus: %v: %v", waitStatus, err) + log.Debugf("containerManager.Wait returned, cid: %s, waitStatus: %#x, err: %v", *cid, *waitStatus, err) return err } @@ -470,8 +469,10 @@ type WaitPIDArgs struct { // WaitPID waits for the process with PID 'pid' in the sandbox. func (cm *containerManager) WaitPID(args *WaitPIDArgs, waitStatus *uint32) error { - log.Debugf("containerManager.Wait") - return cm.l.waitPID(kernel.ThreadID(args.PID), args.CID, waitStatus) + log.Debugf("containerManager.Wait, cid: %s, pid: %d", args.CID, args.PID) + err := cm.l.waitPID(kernel.ThreadID(args.PID), args.CID, waitStatus) + log.Debugf("containerManager.Wait, cid: %s, pid: %d, waitStatus: %#x, err: %v", args.CID, args.PID, *waitStatus, err) + return err } // SignalDeliveryMode enumerates different signal delivery modes. @@ -528,6 +529,6 @@ type SignalArgs struct { // indicated process, to all processes in the container, or to the foreground // process group. func (cm *containerManager) Signal(args *SignalArgs, _ *struct{}) error { - log.Debugf("containerManager.Signal %+v", args) + log.Debugf("containerManager.Signal: cid: %s, PID: %d, signal: %d, mode: %v", args.CID, args.PID, args.Signo, args.Mode) return cm.l.signal(args.CID, args.PID, args.Signo, args.Mode) } diff --git a/runsc/boot/loader.go b/runsc/boot/loader.go index db2b0cea7..ebdd518d0 100644 --- a/runsc/boot/loader.go +++ b/runsc/boot/loader.go @@ -888,7 +888,7 @@ func (l *Loader) destroyContainer(cid string) error { } } - log.Debugf("Container destroyed %q", cid) + log.Debugf("Container destroyed, cid: %s", cid) return nil } |