Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 14 additions & 2 deletions cgroup/cgroup.go
Original file line number Diff line number Diff line change
Expand Up @@ -272,7 +272,7 @@ func reconcileFence(parentDir, fence string) error {
current, err := readControl(parentDir, cpusetName)
if fence == "" {
if err == nil && current != "" {
return writeControl(parentDir, cpusetName, "\n")
return explainFenceWrite(parentDir, current, fence, writeControl(parentDir, cpusetName, "\n"))
}
return nil
}
Expand All @@ -288,7 +288,7 @@ func reconcileFence(parentDir, fence string) error {
if err := checkScopePlacements(parentDir, fence); err != nil {
return err
}
return writeControl(parentDir, cpusetName, fence)
return explainFenceWrite(parentDir, current, fence, writeControl(parentDir, cpusetName, fence))
}

// placeScope applies a per-VM placement; the kernel silently degrades ungrantable requests to the parent set, so the subset check is cocoon's.
Expand Down Expand Up @@ -355,6 +355,18 @@ func checkScopePlacements(parentDir, fence string) error {
return nil
}

// explainFenceWrite names the live VMs behind a rejected fence write: the kernel refuses to re-home populated cgroups and reports only ENOSPC, which reads as an unrelated disk-full error.
func explainFenceWrite(parentDir, current, fence string, err error) error {
if err == nil {
return nil
}
ids, listErr := ListScopeVMIDs(parentDir)
if listErr != nil || len(ids) == 0 {
return err
}
return fmt.Errorf("cgroup_cpus %q -> %q rejected with %d VM(s) running; the fence is machine-wide, drain them first: %w", current, fence, len(ids), err)
}

func forEachLevel(rel string, fn func(dir string) error) error {
if err := fn(Root); err != nil {
return err
Expand Down
24 changes: 24 additions & 0 deletions cgroup/cgroup_test.go
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
package cgroup

import (
"errors"
"os"
"path/filepath"
"slices"
Expand Down Expand Up @@ -210,6 +211,29 @@ func TestReconcileFenceCanonicalEquality(t *testing.T) {
}
}

func TestExplainFenceWriteNamesLiveVMs(t *testing.T) {
parent := t.TempDir()
enospc := errors.New("no space left on device")

if got := explainFenceWrite(parent, "0-14", "", enospc); got != enospc {
t.Errorf("no live scope: got %v, want the error unchanged", got)
}
if got := explainFenceWrite(parent, "0-14", "", nil); got != nil {
t.Errorf("successful write: got %v, want nil", got)
}

if err := os.Mkdir(ScopeDir(parent, "X"), 0o755); err != nil {
t.Fatalf("setup: %v", err)
}
got := explainFenceWrite(parent, "0-14", "", enospc)
if !errors.Is(got, enospc) {
t.Fatalf("wrapped error dropped the cause: %v", got)
}
if !strings.Contains(got.Error(), "1 VM(s) running") {
t.Errorf("message does not name the live VMs: %v", got)
}
}

func TestPlaceScopeReadGate(t *testing.T) {
parent := t.TempDir()
dir := ScopeDir(parent, "X")
Expand Down