Merge branch 'develop' into develop

This commit is contained in:
暗音
2019-10-28 11:06:13 +08:00
committed by GitHub
18 changed files with 364 additions and 228 deletions

View File

@@ -24,7 +24,7 @@ func GetLocalLog(logPath string) (fileBytes []byte, err error) {
}
defer utils.Close(f)
const bufLen = 2 * 1024 * 1024
const bufLen = 1 * 1024 * 1024
logBuf := make([]byte, bufLen)
off := int64(0)

View File

@@ -4,6 +4,7 @@ import (
"crawlab/constants"
"crawlab/database"
"crawlab/services/register"
"errors"
"github.com/apex/log"
"github.com/globalsign/mgo"
"github.com/globalsign/mgo/bson"
@@ -156,15 +157,20 @@ func GetNodeList(filter interface{}) ([]Node, error) {
}
func GetNode(id bson.ObjectId) (Node, error) {
var node Node
if id.Hex() == "" {
log.Infof("id is empty")
debug.PrintStack()
return node, errors.New("id is empty")
}
s, c := database.GetCol("nodes")
defer s.Close()
var node Node
if err := c.FindId(id).One(&node); err != nil {
if err != mgo.ErrNotFound {
log.Errorf(err.Error())
debug.PrintStack()
}
log.Errorf(err.Error())
debug.PrintStack()
return node, err
}
return node, nil

View File

@@ -16,6 +16,7 @@ type Schedule struct {
Description string `json:"description" bson:"description"`
SpiderId bson.ObjectId `json:"spider_id" bson:"spider_id"`
NodeId bson.ObjectId `json:"node_id" bson:"node_id"`
NodeKey string `json:"node_key" bson:"node_key"`
Cron string `json:"cron" bson:"cron"`
EntryId cron.EntryID `json:"entry_id" bson:"entry_id"`
Param string `json:"param" bson:"param"`
@@ -38,6 +39,33 @@ func (sch *Schedule) Save() error {
return nil
}
func (sch *Schedule) Delete() error {
s, c := database.GetCol("schedules")
defer s.Close()
return c.RemoveId(sch.Id)
}
func (sch *Schedule) SyncNodeIdAndSpiderId(node Node, spider Spider) {
sch.syncNodeId(node)
sch.syncSpiderId(spider)
}
func (sch *Schedule) syncNodeId(node Node) {
if node.Id.Hex() == sch.NodeId.Hex() {
return
}
sch.NodeId = node.Id
_ = sch.Save()
}
func (sch *Schedule) syncSpiderId(spider Spider) {
if spider.Id.Hex() == sch.SpiderId.Hex() {
return
}
sch.SpiderId = spider.Id
_ = sch.Save()
}
func GetScheduleList(filter interface{}) ([]Schedule, error) {
s, c := database.GetCol("schedules")
defer s.Close()
@@ -47,11 +75,12 @@ func GetScheduleList(filter interface{}) ([]Schedule, error) {
return schedules, err
}
for i, schedule := range schedules {
var schs []Schedule
for _, schedule := range schedules {
// 获取节点名称
if schedule.NodeId == bson.ObjectIdHex(constants.ObjectIdNull) {
// 选择所有节点
schedules[i].NodeName = "All Nodes"
schedule.NodeName = "All Nodes"
} else {
// 选择单一节点
node, err := GetNode(schedule.NodeId)
@@ -59,7 +88,7 @@ func GetScheduleList(filter interface{}) ([]Schedule, error) {
log.Errorf(err.Error())
continue
}
schedules[i].NodeName = node.Name
schedule.NodeName = node.Name
}
// 获取爬虫名称
@@ -67,11 +96,13 @@ func GetScheduleList(filter interface{}) ([]Schedule, error) {
if err != nil {
log.Errorf("get spider by id: %s, error: %s", schedule.SpiderId.Hex(), err.Error())
debug.PrintStack()
_ = schedule.Delete()
continue
}
schedules[i].SpiderName = spider.Name
schedule.SpiderName = spider.Name
schs = append(schs, schedule)
}
return schedules, nil
return schs, nil
}
func GetSchedule(id bson.ObjectId) (Schedule, error) {
@@ -93,7 +124,12 @@ func UpdateSchedule(id bson.ObjectId, item Schedule) error {
if err := c.FindId(id).One(&result); err != nil {
return err
}
node, err := GetNode(item.NodeId)
if err != nil {
return err
}
item.NodeKey = node.Key
if err := item.Save(); err != nil {
return err
}
@@ -104,9 +140,15 @@ func AddSchedule(item Schedule) error {
s, c := database.GetCol("schedules")
defer s.Close()
node, err := GetNode(item.NodeId)
if err != nil {
return err
}
item.Id = bson.NewObjectId()
item.CreateTs = time.Now()
item.UpdateTs = time.Now()
item.NodeKey = node.Key
if err := c.Insert(&item); err != nil {
debug.PrintStack()

View File

@@ -98,6 +98,12 @@ func (spider *Spider) GetLastTask() (Task, error) {
return tasks[0], nil
}
func (spider *Spider) Delete() error {
s, c := database.GetCol("spiders")
defer s.Close()
return c.RemoveId(spider.Id)
}
// 爬虫列表
func GetSpiderList(filter interface{}, skip int, limit int) ([]Spider, int, error) {
s, c := database.GetCol("spiders")

View File

@@ -4,7 +4,6 @@ import (
"crawlab/constants"
"crawlab/database"
"github.com/apex/log"
"github.com/globalsign/mgo"
"github.com/globalsign/mgo/bson"
"runtime/debug"
"time"
@@ -118,20 +117,16 @@ func GetTaskList(filter interface{}, skip int, limit int, sortKey string) ([]Tas
for i, task := range tasks {
// 获取爬虫名称
spider, err := task.GetSpider()
if err == mgo.ErrNotFound {
// do nothing
} else if err != nil {
return tasks, err
if spider.Id.Hex() == "" || err != nil {
_ = spider.Delete()
} else {
tasks[i].SpiderName = spider.DisplayName
}
// 获取节点名称
node, err := task.GetNode()
if err == mgo.ErrNotFound {
// do nothing
} else if err != nil {
return tasks, err
if node.Id.Hex() == "" || err != nil {
_ = task.Delete()
} else {
tasks[i].NodeName = node.Name
}

View File

@@ -15,9 +15,9 @@ func GetNodeList(c *gin.Context) {
return
}
for i, node := range nodes {
nodes[i].IsMaster = services.IsMasterNode(node.Id.Hex())
}
//for i, node := range nodes {
// nodes[i].IsMaster = services.IsMasterNode(node.Id.Hex())
//}
c.JSON(http.StatusOK, Response{
Status: "ok",
@@ -109,11 +109,11 @@ func GetSystemInfo(c *gin.Context) {
})
}
func DeleteNode(c *gin.Context) {
func DeleteNode(c *gin.Context) {
id := c.Param("id")
node, err := model.GetNode(bson.ObjectIdHex(id))
if err != nil {
HandleError(http.StatusInternalServerError, c ,err)
HandleError(http.StatusInternalServerError, c, err)
return
}
err = node.Delete()

View File

@@ -1,7 +1,6 @@
package routes
import (
"crawlab/constants"
"crawlab/model"
"crawlab/services"
"github.com/gin-gonic/gin"
@@ -46,13 +45,14 @@ func PostSchedule(c *gin.Context) {
HandleError(http.StatusBadRequest, c, err)
return
}
newItem.Id = bson.ObjectIdHex(id)
// 如果node_id为空则置为空ObjectId
if newItem.NodeId == "" {
newItem.NodeId = bson.ObjectIdHex(constants.ObjectIdNull)
// 验证cron表达式
if err := services.ParserCron(newItem.Cron); err != nil {
HandleError(http.StatusOK, c, err)
return
}
newItem.Id = bson.ObjectIdHex(id)
// 更新数据库
if err := model.UpdateSchedule(bson.ObjectIdHex(id), newItem); err != nil {
HandleError(http.StatusInternalServerError, c, err)
@@ -80,9 +80,10 @@ func PutSchedule(c *gin.Context) {
return
}
// 如果node_id为空则置为空ObjectId
if item.NodeId == "" {
item.NodeId = bson.ObjectIdHex(constants.ObjectIdNull)
// 验证cron表达式
if err := services.ParserCron(item.Cron); err != nil {
HandleError(http.StatusOK, c, err)
return
}
// 更新数据库

View File

@@ -15,6 +15,7 @@ import (
"os"
"path/filepath"
"runtime/debug"
"time"
)
// 任务日志频道映射
@@ -45,8 +46,14 @@ func GetRemoteLog(task model.Task) (logStr string, err error) {
// 生成频道等待获取log
ch := TaskLogChanMap.ChanBlocked(task.Id)
// 此处阻塞,等待结果
logStr = <-ch
select {
case logStr = <-ch:
log.Infof("get remote log")
break
case <-time.After(30 * time.Second):
logStr = "get remote log timeout"
break
}
return logStr, nil
}

View File

@@ -3,6 +3,7 @@ package msg_handler
import (
"crawlab/constants"
"crawlab/entity"
"github.com/apex/log"
)
type Handler interface {
@@ -10,6 +11,7 @@ type Handler interface {
}
func GetMsgHandler(msg entity.NodeMessage) Handler {
log.Infof("received msg , type is : %s", msg.Type)
if msg.Type == constants.MsgTypeGetLog || msg.Type == constants.MsgTypeRemoveLog {
// 日志相关
return &Log{

View File

@@ -40,9 +40,12 @@ func (g *Log) get() error {
msgSd.Log = utils.BytesToString(logStr)
}
// 发布消息给主节点
if err := database.Pub(constants.ChannelMasterNode, msgSd); err != nil {
if err := utils.Pub(constants.ChannelMasterNode, msgSd); err != nil {
log.Errorf("pub log to master node error: %s", err.Error())
debug.PrintStack()
return err
}
log.Infof(msgSd.Log)
return nil
}

View File

@@ -88,6 +88,8 @@ func UpdateNodeStatus() {
handleNodeInfo(key, data)
}
// 重新获取list
list, _ = database.RedisClient.HKeys("nodes")
// 重置不在redis的key为offline
model.ResetNodeStatusToOffline(list)
}
@@ -110,13 +112,15 @@ func handleNodeInfo(key string, data Data) {
if err := c.Find(bson.M{"key": key}).One(&node); err != nil {
// 数据库不存在该节点
node = model.Node{
Key: key,
Name: data.Ip,
Ip: data.Ip,
Port: "8000",
Mac: data.Mac,
Status: constants.StatusOnline,
IsMaster: data.Master,
Key: key,
Name: data.Ip,
Ip: data.Ip,
Port: "8000",
Mac: data.Mac,
Status: constants.StatusOnline,
IsMaster: data.Master,
UpdateTs: time.Now(),
UpdateTsUnix: time.Now().Unix(),
}
if err := node.Add(); err != nil {
log.Errorf(err.Error())
@@ -125,6 +129,8 @@ func handleNodeInfo(key string, data Data) {
} else {
// 数据库存在该节点
node.Status = constants.StatusOnline
node.UpdateTs = time.Now()
node.UpdateTsUnix = time.Now().Unix()
if err := node.Save(); err != nil {
log.Errorf(err.Error())
return
@@ -205,6 +211,8 @@ func WorkerNodeCallback(message redis.Message) (err error) {
// 反序列化
msg := utils.GetMessage(message)
if err := msg_handler.GetMsgHandler(*msg).Handle(); err != nil {
log.Errorf("msg handler error: %s", err.Error())
debug.PrintStack()
return err
}
return nil

View File

@@ -5,7 +5,7 @@ import (
"crawlab/lib/cron"
"crawlab/model"
"github.com/apex/log"
uuid "github.com/satori/go.uuid"
"github.com/satori/go.uuid"
"runtime/debug"
)
@@ -17,7 +17,22 @@ type Scheduler struct {
func AddTask(s model.Schedule) func() {
return func() {
nodeId := s.NodeId
node, err := model.GetNodeByKey(s.NodeKey)
if err != nil || node.Id.Hex() == "" {
log.Errorf("get node by key error: %s", err.Error())
debug.PrintStack()
return
}
spider := model.GetSpiderByName(s.SpiderName)
if spider == nil || spider.Id.Hex() == "" {
log.Errorf("get spider by name error: %s", err.Error())
debug.PrintStack()
return
}
// 同步ID到定时任务
s.SyncNodeIdAndSpiderId(node, *spider)
// 生成任务ID
id := uuid.NewV4()
@@ -25,8 +40,8 @@ func AddTask(s model.Schedule) func() {
// 生成任务模型
t := model.Task{
Id: id.String(),
SpiderId: s.SpiderId,
NodeId: nodeId,
SpiderId: spider.Id,
NodeId: node.Id,
Status: constants.StatusPending,
Param: s.Param,
}
@@ -107,6 +122,18 @@ func (s *Scheduler) RemoveAll() {
}
}
// 验证cron表达式是否正确
func ParserCron(spec string) error {
parser := cron.NewParser(
cron.Second | cron.Minute | cron.Hour | cron.Dom | cron.Month | cron.Dow | cron.Descriptor,
)
if _, err := parser.Parse(spec); err != nil {
return err
}
return nil
}
func (s *Scheduler) Update() error {
// 删除所有定时任务
s.RemoveAll()

View File

@@ -100,91 +100,85 @@ func AssignTask(task model.Task) error {
return nil
}
// 执行shell命令
func ExecuteShellCmd(cmdStr string, cwd string, t model.Task, s model.Spider) (err error) {
log.Infof("cwd: " + cwd)
log.Infof("cmd: " + cmdStr)
// 设置环境变量
func SetEnv(cmd *exec.Cmd, envs []model.Env, taskId string, dataCol string) *exec.Cmd {
// 默认环境变量
cmd.Env = append(os.Environ(), "CRAWLAB_TASK_ID="+taskId)
cmd.Env = append(cmd.Env, "CRAWLAB_COLLECTION="+dataCol)
cmd.Env = append(cmd.Env, "PYTHONUNBUFFERED=0")
cmd.Env = append(cmd.Env, "PYTHONIOENCODING=utf-8")
cmd.Env = append(cmd.Env, "TZ=Asia/Shanghai")
// 生成执行命令
var cmd *exec.Cmd
if runtime.GOOS == constants.Windows {
cmd = exec.Command("cmd", "/C", cmdStr)
} else {
cmd = exec.Command("sh", "-c", cmdStr)
}
// 工作目录
cmd.Dir = cwd
// 指定stdout, stderr日志位置
fLog, err := os.Create(t.LogPath)
if err != nil {
HandleTaskError(t, err)
return err
}
defer utils.Close(fLog)
cmd.Stdout = fLog
cmd.Stderr = fLog
// 添加默认环境变量
cmd.Env = append(cmd.Env, "CRAWLAB_TASK_ID="+t.Id)
cmd.Env = append(cmd.Env, "CRAWLAB_COLLECTION="+s.Col)
// 添加任务环境变量
for _, env := range s.Envs {
//任务环境变量
for _, env := range envs {
cmd.Env = append(cmd.Env, env.Name+"="+env.Value)
}
// 起一个goroutine来监控进程
ch := utils.TaskExecChanMap.ChanBlocked(t.Id)
go func() {
// 传入信号,此处阻塞
signal := <-ch
log.Infof("cancel process signal: %s", signal)
if signal == constants.TaskCancel && cmd.Process != nil {
// 取消进程
if err := syscall.Kill(-cmd.Process.Pid, syscall.SIGKILL); err != nil {
log.Errorf("process kill error: %s", err.Error())
debug.PrintStack()
}
t.Status = constants.StatusCancelled
} else {
// 保存任务
t.Status = constants.StatusFinished
}
t.FinishTs = time.Now()
t.Error = "user kill the process ..."
if err := t.Save(); err != nil {
log.Infof("save task error: %s", err.Error())
// TODO 全局环境变量
return cmd
}
func SetLogConfig(cmd *exec.Cmd, path string) error {
fLog, err := os.Create(path)
if err != nil {
log.Errorf("create task log file error: %s", path)
debug.PrintStack()
return err
}
cmd.Stdout = fLog
cmd.Stderr = fLog
return nil
}
func FinishOrCancelTask(ch chan string, cmd *exec.Cmd, t model.Task) {
// 传入信号,此处阻塞
signal := <-ch
log.Infof("process received signal: %s", signal)
if signal == constants.TaskCancel && cmd.Process != nil {
// 取消进程
if err := syscall.Kill(-cmd.Process.Pid, syscall.SIGKILL); err != nil {
log.Errorf("process kill error: %s", err.Error())
debug.PrintStack()
return
t.Error = "kill process error: " + err.Error()
t.Status = constants.StatusError
} else {
t.Error = "user kill the process ..."
t.Status = constants.StatusCancelled
}
}()
} else {
// 保存任务
t.Status = constants.StatusFinished
}
// 在选择所有节点执行的时候,实际就是随机一个节点执行的,
cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
t.FinishTs = time.Now()
_ = t.Save()
}
// 异步启动进程
func StartTaskProcess(cmd *exec.Cmd, t model.Task) error {
if err := cmd.Start(); err != nil {
log.Errorf("start spider error:{}", err.Error())
debug.PrintStack()
return err
}
// 保存pid到task
t.Pid = cmd.Process.Pid
if err := t.Save(); err != nil {
log.Errorf("save task pid error: %s", err.Error())
debug.PrintStack()
t.Error = "start task error: " + err.Error()
t.Status = constants.StatusError
t.FinishTs = time.Now()
_ = t.Save()
return err
}
// 同步等待进程完成
return nil
}
func WaitTaskProcess(cmd *exec.Cmd, t model.Task) error {
if err := cmd.Wait(); err != nil {
log.Errorf("wait process finish error: %s", err.Error())
debug.PrintStack()
if exitError, ok := err.(*exec.ExitError); ok {
exitCode := exitError.ExitCode()
log.Errorf("exit error, exit code: %d", exitCode)
// 非kill 的错误类型
if exitCode != -1 {
// 非手动kill保存为错误状态
@@ -194,6 +188,52 @@ func ExecuteShellCmd(cmdStr string, cwd string, t model.Task, s model.Spider) (e
_ = t.Save()
}
}
return err
}
return nil
}
// 执行shell命令
func ExecuteShellCmd(cmdStr string, cwd string, t model.Task, s model.Spider) (err error) {
log.Infof("cwd: %s", cwd)
log.Infof("cmd: %s", cmdStr)
// 生成执行命令
var cmd *exec.Cmd
if runtime.GOOS == constants.Windows {
cmd = exec.Command("cmd", "/C", cmdStr)
} else {
cmd = exec.Command("")
cmd = exec.Command("sh", "-c", cmdStr)
}
// 工作目录
cmd.Dir = cwd
// 日志配置
if err := SetLogConfig(cmd, t.LogPath); err != nil {
return err
}
// 环境变量配置
cmd = SetEnv(cmd, s.Envs, t.Id, s.Col)
// 起一个goroutine来监控进程
ch := utils.TaskExecChanMap.ChanBlocked(t.Id)
go FinishOrCancelTask(ch, cmd, t)
// kill的时候可以kill所有的子进程
cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
// 启动进程
if err := StartTaskProcess(cmd, t); err != nil {
return err
}
// 同步等待进程完成
if err := WaitTaskProcess(cmd, t); err != nil {
return err
}
ch <- constants.TaskFinish
@@ -208,6 +248,7 @@ func MakeLogDir(t model.Task) (fileDir string, err error) {
// 如果日志目录不存在,生成该目录
if !utils.Exists(fileDir) {
if err := os.MkdirAll(fileDir, 0777); err != nil {
log.Errorf("execute task, make log dir error: %s", err.Error())
debug.PrintStack()
return "", err
}
@@ -272,82 +313,55 @@ func ExecuteTask(id int) {
// 获取当前节点
node, err := model.GetCurrentNode()
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
log.Errorf("execute task get current node error: %s", err.Error())
debug.PrintStack()
return
}
// 公共队列
queuePub := "tasks:public"
// 节点队列
queueCur := "tasks:node:" + node.Id.Hex()
// 节点队列任务
var msg string
msg, err = database.RedisClient.LPop(queueCur)
if err != nil {
if msg == "" {
// 节点队列没有任务,获取公共队列任务
msg, err = database.RedisClient.LPop(queuePub)
if err != nil {
if msg == "" {
// 公共队列没有任务
log.Debugf(GetWorkerPrefix(id) + "没有任务...")
return
} else {
log.Errorf(GetWorkerPrefix(id) + err.Error())
debug.PrintStack()
return
}
}
} else {
log.Errorf(GetWorkerPrefix(id) + err.Error())
debug.PrintStack()
return
if msg, err = database.RedisClient.LPop(queueCur); err != nil {
// 节点队列没有任务,获取公共队列任务
queuePub := "tasks:public"
if msg, err = database.RedisClient.LPop(queuePub); err != nil {
}
}
if msg == "" {
return
}
// 反序列化
tMsg := TaskMessage{}
if err := json.Unmarshal([]byte(msg), &tMsg); err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
debug.PrintStack()
log.Errorf("json string to struct error: %s", err.Error())
return
}
// 获取任务
t, err := model.GetTask(tMsg.Id)
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
log.Errorf("execute task, get task error: %s", err.Error())
return
}
// 获取爬虫
spider, err := t.GetSpider()
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
log.Errorf("execute task, get spider error: %s", err.Error())
return
}
// 创建日志目录
fileDir, err := MakeLogDir(t)
if err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
var fileDir string
if fileDir, err = MakeLogDir(t); err != nil {
return
}
// 获取日志文件路径
t.LogPath = GetLogFilePaths(fileDir)
// 创建日志目录文件夹
fileStdoutDir := filepath.Dir(t.LogPath)
if !utils.Exists(fileStdoutDir) {
if err := os.MkdirAll(fileStdoutDir, os.ModePerm); err != nil {
log.Errorf(GetWorkerPrefix(id) + err.Error())
return
}
}
// 工作目录
cwd := filepath.Join(
viper.GetString("spider.path"),