diff --git a/node-tools/internal/launch/launch.go b/node-tools/internal/launch/launch.go index 0044ea5..bcb593c 100644 --- a/node-tools/internal/launch/launch.go +++ b/node-tools/internal/launch/launch.go @@ -345,6 +345,7 @@ func Start(module, entry string, env []string, mesh Bus, logf func(string, ...an go func(m message) { var result any = map[string]any{} var err error + asked := time.Now() switch m.Method { case "mesh/publish": err = mesh.Publish(m.Params) @@ -479,6 +480,13 @@ func Start(module, entry string, env []string, mesh Bus, logf func(string, ...an } return } + // A state or log operation the bus refused or let time out reaches a caller as a call that + // does not answer: say it here, with how long it took, so the next hang is readable. + if err != nil && (strings.HasPrefix(m.Method, "mesh/state.") || strings.HasPrefix(m.Method, "mesh/log.") || + busStateCalls[m.Method]) { + logf("[mesh-tools] %s's bundle asked %s and it failed after %s: %v", module, m.Method, + time.Since(asked).Round(time.Millisecond), err) + } if len(m.ID) == 0 { return } diff --git a/node-tools/internal/launch/state_failed_test.go b/node-tools/internal/launch/state_failed_test.go new file mode 100644 index 0000000..7ee0f99 --- /dev/null +++ b/node-tools/internal/launch/state_failed_test.go @@ -0,0 +1,64 @@ +package launch + +import ( + "encoding/json" + "errors" + "fmt" + "os" + "path/filepath" + "strings" + "sync" + "testing" + "time" +) + +type refusingState struct{ subscribing } + +func (*refusingState) State(string, json.RawMessage) (json.RawMessage, error) { + return nil, errors.New("the bus did not answer in 10s") +} + +// A bundle that asks the runtime for a state once it is started, as a tool does. +const askingBundle = `#!/bin/sh +while IFS= read -r line; do + id=$(printf '%s' "$line" | sed -n 's/.*"id":\([0-9]*\)[,}].*/\1/p') + case "$line" in + *'"method":"initialize"'*) printf '{"jsonrpc":"2.0","id":%s,"result":{}}\n' "$id" ;; + *'"method":"tools/list"'*) printf '{"jsonrpc":"2.0","id":%s,"result":{"tools":[]}}\n' "$id" + printf '{"jsonrpc":"2.0","id":"q1","method":"mesh/state.get","params":{"state":"copied","key":"k"}}\n' ;; + esac +done +` + +// A state operation the bus refuses or lets time out is said in the runtime's log with its method, its bundle and its +// time (issue 1966: a hand-act seat whose calls hung said nothing). +func TestAFailedStateOperationIsSaidInTheRuntimesLog(t *testing.T) { + dir := t.TempDir() + entry := filepath.Join(dir, "bundle") + if err := os.WriteFile(entry, []byte(askingBundle), 0o755); err != nil { + t.Fatal(err) + } + var mu sync.Mutex + var lines []string + logf := func(f string, a ...any) { + mu.Lock() + defer mu.Unlock() + lines = append(lines, fmt.Sprintf(f, a...)) + } + l, err := Start("asker", entry, os.Environ(), &refusingState{}, logf) + if err != nil { + t.Fatal(err) + } + t.Cleanup(l.Stop) + deadline := time.Now().Add(5 * time.Second) + for time.Now().Before(deadline) { + mu.Lock() + all := strings.Join(lines, "\n") + mu.Unlock() + if strings.Contains(all, "asker's bundle asked mesh/state.get and it failed after") && strings.Contains(all, "did not answer in 10s") { + return + } + time.Sleep(20 * time.Millisecond) + } + t.Fatalf("the failed state operation was not said; the log: %v", lines) +}