podlogs.go

/*
Copyright 2018 The Kubernetes Authors.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/

// Package podlogs enables live capturing of all events and log
// messages for some or all pods in a namespace as they get generated.
// This helps debugging both a running test (what is currently going
// on?) and the output of a CI run (events appear in chronological
// order and output that normally isn't available like the command
// stdout messages are available).
package podlogs

import (
	"bufio"
	"bytes"
	"context"
	"fmt"
	"github.com/pkg/errors"
	"io"
	"os"
	"path"
	"regexp"
	"strings"
	"sync"

	"k8s.io/api/core/v1"
	meta "k8s.io/apimachinery/pkg/apis/meta/v1"
	clientset "k8s.io/client-go/kubernetes"
)

// LogsForPod starts reading the logs for a certain pod. If the pod has more than one
// container, opts.Container must be set. Reading stops when the context is done.
// The stream includes formatted error messages and ends with
//    rpc error: code = Unknown desc = Error: No such container: 41a...
// when the pod gets deleted while streaming.
func LogsForPod(ctx context.Context, cs clientset.Interface, ns, pod string, opts *v1.PodLogOptions) (io.ReadCloser, error) {
	req := cs.Core().Pods(ns).GetLogs(pod, opts)
	return req.Context(ctx).Stream()
}

// LogOutput determines where output from CopyAllLogs goes.
type LogOutput struct {
	// If not nil, errors will be logged here.
	StatusWriter io.Writer

	// If not nil, all output goes to this writer with "<pod>/<container>:" as prefix.
	LogWriter io.Writer

	// Base directory for one log file per container.
	// The full path of each log file will be <log path prefix><pod>-<container>.log.
	LogPathPrefix string
}

// Matches harmless errors from pkg/kubelet/kubelet_pods.go.
var expectedErrors = regexp.MustCompile(`container .* in pod .* is (terminated|waiting to start|not available)|the server could not find the requested resource`)

// CopyAllLogs follows the logs of all containers in all pods,
// including those that get created in the future, and writes each log
// line as configured in the output options. It does that until the
// context is done or until an error occurs.
//
// Beware that there is currently no way to force log collection
// before removing pods, which means that there is a known race
// between "stop pod" and "collecting log entries". The alternative
// would be a blocking function with collects logs from all currently
// running pods, but that then would have the disadvantage that
// already deleted pods aren't covered.
func CopyAllLogs(ctx context.Context, cs clientset.Interface, ns string, to LogOutput) error {
	watcher, err := cs.Core().Pods(ns).Watch(meta.ListOptions{})
	if err != nil {
		return errors.Wrap(err, "cannot create Pod event watcher")
	}

	go func() {
		var m sync.Mutex
		logging := map[string]bool{}
		check := func() {
			m.Lock()
			defer m.Unlock()

			pods, err := cs.Core().Pods(ns).List(meta.ListOptions{})
			if err != nil {
				if to.StatusWriter != nil {
					fmt.Fprintf(to.StatusWriter, "ERROR: get pod list in %s: %s\n", ns, err)
				}
				return
			}

			for _, pod := range pods.Items {
				for _, c := range pod.Spec.Containers {
					name := pod.ObjectMeta.Name + "/" + c.Name
					if logging[name] {
						continue
					}
					readCloser, err := LogsForPod(ctx, cs, ns, pod.ObjectMeta.Name,
						&v1.PodLogOptions{
							Container: c.Name,
							Follow:    true,
						})
					if err != nil {
						// We do get "normal" errors here, like trying to read too early.
						// We can ignore those.
						if to.StatusWriter != nil &&
							expectedErrors.FindStringIndex(err.Error()) == nil {
							fmt.Fprintf(to.StatusWriter, "WARNING: pod log: %s: %s\n", name, err)
						}
						continue
					}

					// Determine where we write. If this fails, we intentionally return without clearing
					// the logging[name] flag, which prevents trying over and over again to
					// create the output file.
					var out io.Writer
					var closer io.Closer
					var prefix string
					if to.LogWriter != nil {
						out = to.LogWriter
						prefix = name + ": "
					} else {
						var err error
						filename := to.LogPathPrefix + pod.ObjectMeta.Name + "-" + c.Name + ".log"
						err = os.MkdirAll(path.Dir(filename), 0755)
						if err != nil {
							if to.StatusWriter != nil {
								fmt.Fprintf(to.StatusWriter, "ERROR: pod log: create directory for %s: %s\n", filename, err)
							}
							return
						}
						// The test suite might run the same test multiple times,
						// so we have to append here.
						file, err := os.OpenFile(filename, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0644)
						if err != nil {
							if to.StatusWriter != nil {
								fmt.Fprintf(to.StatusWriter, "ERROR: pod log: create file %s: %s\n", filename, err)
							}
							return
						}
						closer = file
						out = file
					}
					go func() {
						if closer != nil {
							defer closer.Close()
						}
						defer func() {
							m.Lock()
							logging[name] = false
							m.Unlock()
							readCloser.Close()
						}()
						scanner := bufio.NewScanner(readCloser)
						first := true
						for scanner.Scan() {
							line := scanner.Text()
							// Filter out the expected "end of stream" error message,
							// it would just confuse developers who don't know about it.
							// Same for attempts to read logs from a container that
							// isn't ready (yet?!).
							if !strings.HasPrefix(line, "rpc error: code = Unknown desc = Error: No such container:") &&
								!strings.HasPrefix(line, "Unable to retrieve container logs for ") {
								if first {
									if to.LogWriter == nil {
										// Because the same log might be written to multiple times
										// in different test instances, log an extra line to separate them.
										// Also provides some useful extra information.
										fmt.Fprintf(out, "==== start of log for container %s ====\n", name)
									}
									first = false
								}
								fmt.Fprintf(out, "%s%s\n", prefix, scanner.Text())
							}
						}
					}()
					logging[name] = true
				}
			}
		}

		// Watch events to see whether we can start logging
		// and log interesting ones.
		check()
		for {
			select {
			case <-watcher.ResultChan():
				check()
			case <-ctx.Done():
				return
			}
		}
	}()

	return nil
}

// WatchPods prints pod status events for a certain namespace or all namespaces
// when namespace name is empty.
func WatchPods(ctx context.Context, cs clientset.Interface, ns string, to io.Writer) error {
	watcher, err := cs.Core().Pods(ns).Watch(meta.ListOptions{})
	if err != nil {
		return errors.Wrap(err, "cannot create Pod event watcher")
	}

	go func() {
		defer watcher.Stop()
		for {
			select {
			case e := <-watcher.ResultChan():
				if e.Object == nil {
					continue
				}

				pod, ok := e.Object.(*v1.Pod)
				if !ok {
					continue
				}
				buffer := new(bytes.Buffer)
				fmt.Fprintf(buffer,
					"pod event: %s: %s/%s %s: %s %s\n",
					e.Type,
					pod.Namespace,
					pod.Name,
					pod.Status.Phase,
					pod.Status.Reason,
					pod.Status.Conditions,
				)
				for _, cst := range pod.Status.ContainerStatuses {
					fmt.Fprintf(buffer, "   %s: ", cst.Name)
					if cst.State.Waiting != nil {
						fmt.Fprintf(buffer, "WAITING: %s - %s",
							cst.State.Waiting.Reason,
							cst.State.Waiting.Message,
						)
					} else if cst.State.Running != nil {
						fmt.Fprintf(buffer, "RUNNING")
					} else if cst.State.Waiting != nil {
						fmt.Fprintf(buffer, "TERMINATED: %s - %s",
							cst.State.Waiting.Reason,
							cst.State.Waiting.Message,
						)
					}
					fmt.Fprintf(buffer, "\n")
				}
				to.Write(buffer.Bytes())
			case <-ctx.Done():
				return
			}
		}
	}()

	return nil
}