Reconfiguration scheduling #11
Conversation
| // Clear running jobs | ||
| r.mu.Lock() | ||
| r.runningJobs = make(map[int64]*jobtypes.Job) | ||
| r.mu.Unlock() |
There was a problem hiding this comment.
go 中锁的写法一般在函数开头
使用 defer 释放资源，defer 总是在退出函数时执行
func (r *Runner) Close() error {
r.mu.Lock()
defer r.mu.Unlock()
r.runningJobs = make(map[int64]*jobtypes.Job)
}
There was a problem hiding this comment.
Ok, I will correct this problem. Thank you for your suggestion!
There was a problem hiding this comment.
我觉得这个文件不用删掉，项目确实得有一个地方入口以供测试功能（单测除外
|
|
||
| func startRunners(ctx context.Context, cfg *clrServer.Server) error { | ||
|
|
||
| // Initialize all collectors before starting runners |
There was a problem hiding this comment.
我觉得可以将这部分移动到 job 内部完成 这样项目整体结构会更好理解一些
从 server 启动依赖组件，collector task 也是 job 的一部分？你怎么看 @TJxiaobao
| "hertzbeat.apache.org/hertzbeat-collector-go/internal/util/logger" | ||
| ) | ||
|
|
||
| // init 函数在包被导入时自动执行 |
|
|
||
| // init 函数在包被导入时自动执行 | ||
| // 集中注册所有协议的工厂函数 | ||
| func init() { |
There was a problem hiding this comment.
这里可能仍然有一些出入，不过问题不大。这种方式也挺好的，显式配置更好
| "fmt" | ||
| "time" | ||
|
|
||
| _ "hertzbeat.apache.org/hertzbeat-collector-go/internal/collector/basic" |
There was a problem hiding this comment.
这里是预期中的吗？
它会执行 bacis 包下的所有 init 函数
| } | ||
|
|
||
| // NewResultHandler creates a new result handler | ||
| func NewResultHandler(logger logger.Logger) *ResultHandlerImpl { |
There was a problem hiding this comment.
如果这里是实现了一个接口的话，我想这里应该返回的是接口类型，而不是结构体的实现
| // 4. Triggering alerts based on thresholds | ||
| // 5. Updating monitoring status | ||
|
|
||
| if data.Code == 200 { |
| } | ||
|
|
||
| // NewCommonDispatcher creates a new common dispatcher | ||
| func NewCommonDispatcher(logger logger.Logger, metricsCollector MetricsCollector, resultHandler ResultHandler) *CommonDispatcherImpl { |
There was a problem hiding this comment.
这里应该创建一个 ctx 吗？
从 go 的编程风格来说，ctx 应该永远被用作函数的第一个参数实现，而不是放到结构体里面去，这不符合 go 的编程规范（约定俗成
| commonDispatcher MetricsTaskDispatcher | ||
| cyclicTasks sync.Map // map[int64]*jobtypes.Timeout for cyclic jobs | ||
| tempTasks sync.Map // map[int64]*jobtypes.Timeout for one-time jobs | ||
| ctx context.Context |
| ) | ||
|
|
||
| // hashedWheelTimer implements a time wheel for efficient timeout management | ||
| type hashedWheelTimer struct { |
| bucket.mu.RUnlock() | ||
| } | ||
|
|
||
| return map[string]interface{}{ |
| if err := r.timeDispatch.AddJob(job); err != nil { | ||
| delete(r.runningJobs, job.ID) | ||
| r.Logger.Error(err, "failed to add job to time dispatcher", "jobID", job.ID) | ||
| return fmt.Errorf("failed to add job to time dispatcher: %w", err) |
| // Runner implements the service runner interface | ||
| type Runner struct { | ||
| Config | ||
| timeDispatch TimeDispatcher |
There was a problem hiding this comment.
这些配置应该都在 Config 中，config 是当前 server 的所有 ctx 配置，runner 只负责将当前 server run 起来
3 participants
Based on the existing project framework, refactor the scheduling module of hertzbeat-collector-go.